ioctl.c 136.5 KB
Newer Older
C
Christoph Hellwig 已提交
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23
/*
 * Copyright (C) 2007 Oracle.  All rights reserved.
 *
 * This program is free software; you can redistribute it and/or
 * modify it under the terms of the GNU General Public
 * License v2 as published by the Free Software Foundation.
 *
 * This program is distributed in the hope that it will be useful,
 * but WITHOUT ANY WARRANTY; without even the implied warranty of
 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
 * General Public License for more details.
 *
 * You should have received a copy of the GNU General Public
 * License along with this program; if not, write to the
 * Free Software Foundation, Inc., 59 Temple Place - Suite 330,
 * Boston, MA 021110-1307, USA.
 */

#include <linux/kernel.h>
#include <linux/bio.h>
#include <linux/buffer_head.h>
#include <linux/file.h>
#include <linux/fs.h>
24
#include <linux/fsnotify.h>
C
Christoph Hellwig 已提交
25 26 27 28 29 30
#include <linux/pagemap.h>
#include <linux/highmem.h>
#include <linux/time.h>
#include <linux/init.h>
#include <linux/string.h>
#include <linux/backing-dev.h>
31
#include <linux/mount.h>
C
Christoph Hellwig 已提交
32
#include <linux/mpage.h>
33
#include <linux/namei.h>
C
Christoph Hellwig 已提交
34 35 36 37
#include <linux/swap.h>
#include <linux/writeback.h>
#include <linux/compat.h>
#include <linux/bit_spinlock.h>
38
#include <linux/security.h>
C
Christoph Hellwig 已提交
39
#include <linux/xattr.h>
40
#include <linux/mm.h>
41
#include <linux/slab.h>
42
#include <linux/blkdev.h>
43
#include <linux/uuid.h>
44
#include <linux/btrfs.h>
M
Mark Fasheh 已提交
45
#include <linux/uaccess.h>
C
Christoph Hellwig 已提交
46 47 48 49 50 51
#include "ctree.h"
#include "disk-io.h"
#include "transaction.h"
#include "btrfs_inode.h"
#include "print-tree.h"
#include "volumes.h"
52
#include "locking.h"
53
#include "inode-map.h"
54
#include "backref.h"
55
#include "rcu-string.h"
56
#include "send.h"
57
#include "dev-replace.h"
58
#include "props.h"
59
#include "sysfs.h"
J
Josef Bacik 已提交
60
#include "qgroup.h"
61
#include "tree-log.h"
62
#include "compression.h"
C
Christoph Hellwig 已提交
63

64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89
#ifdef CONFIG_64BIT
/* If we have a 32-bit userspace and 64-bit kernel, then the UAPI
 * structures are incorrect, as the timespec structure from userspace
 * is 4 bytes too small. We define these alternatives here to teach
 * the kernel about the 32-bit struct packing.
 */
struct btrfs_ioctl_timespec_32 {
	__u64 sec;
	__u32 nsec;
} __attribute__ ((__packed__));

struct btrfs_ioctl_received_subvol_args_32 {
	char	uuid[BTRFS_UUID_SIZE];	/* in */
	__u64	stransid;		/* in */
	__u64	rtransid;		/* out */
	struct btrfs_ioctl_timespec_32 stime; /* in */
	struct btrfs_ioctl_timespec_32 rtime; /* out */
	__u64	flags;			/* in */
	__u64	reserved[16];		/* in */
} __attribute__ ((__packed__));

#define BTRFS_IOC_SET_RECEIVED_SUBVOL_32 _IOWR(BTRFS_IOCTL_MAGIC, 37, \
				struct btrfs_ioctl_received_subvol_args_32)
#endif


M
Mark Fasheh 已提交
90
static int btrfs_clone(struct inode *src, struct inode *inode,
91 92
		       u64 off, u64 olen, u64 olen_aligned, u64 destoff,
		       int no_time_update);
M
Mark Fasheh 已提交
93

94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123
/* Mask out flags that are inappropriate for the given type of inode. */
static inline __u32 btrfs_mask_flags(umode_t mode, __u32 flags)
{
	if (S_ISDIR(mode))
		return flags;
	else if (S_ISREG(mode))
		return flags & ~FS_DIRSYNC_FL;
	else
		return flags & (FS_NODUMP_FL | FS_NOATIME_FL);
}

/*
 * Export inode flags to the format expected by the FS_IOC_GETFLAGS ioctl.
 */
static unsigned int btrfs_flags_to_ioctl(unsigned int flags)
{
	unsigned int iflags = 0;

	if (flags & BTRFS_INODE_SYNC)
		iflags |= FS_SYNC_FL;
	if (flags & BTRFS_INODE_IMMUTABLE)
		iflags |= FS_IMMUTABLE_FL;
	if (flags & BTRFS_INODE_APPEND)
		iflags |= FS_APPEND_FL;
	if (flags & BTRFS_INODE_NODUMP)
		iflags |= FS_NODUMP_FL;
	if (flags & BTRFS_INODE_NOATIME)
		iflags |= FS_NOATIME_FL;
	if (flags & BTRFS_INODE_DIRSYNC)
		iflags |= FS_DIRSYNC_FL;
L
Li Zefan 已提交
124 125 126
	if (flags & BTRFS_INODE_NODATACOW)
		iflags |= FS_NOCOW_FL;

127
	if (flags & BTRFS_INODE_NOCOMPRESS)
L
Li Zefan 已提交
128
		iflags |= FS_NOCOMP_FL;
129 130
	else if (flags & BTRFS_INODE_COMPRESS)
		iflags |= FS_COMPR_FL;
131 132 133 134 135 136 137 138 139 140

	return iflags;
}

/*
 * Update inode->i_flags based on the btrfs internal flags.
 */
void btrfs_update_iflags(struct inode *inode)
{
	struct btrfs_inode *ip = BTRFS_I(inode);
141
	unsigned int new_fl = 0;
142 143

	if (ip->flags & BTRFS_INODE_SYNC)
144
		new_fl |= S_SYNC;
145
	if (ip->flags & BTRFS_INODE_IMMUTABLE)
146
		new_fl |= S_IMMUTABLE;
147
	if (ip->flags & BTRFS_INODE_APPEND)
148
		new_fl |= S_APPEND;
149
	if (ip->flags & BTRFS_INODE_NOATIME)
150
		new_fl |= S_NOATIME;
151
	if (ip->flags & BTRFS_INODE_DIRSYNC)
152 153 154 155 156
		new_fl |= S_DIRSYNC;

	set_mask_bits(&inode->i_flags,
		      S_SYNC | S_APPEND | S_IMMUTABLE | S_NOATIME | S_DIRSYNC,
		      new_fl);
157 158 159 160
}

static int btrfs_ioctl_getflags(struct file *file, void __user *arg)
{
A
Al Viro 已提交
161
	struct btrfs_inode *ip = BTRFS_I(file_inode(file));
162 163 164 165 166 167 168
	unsigned int flags = btrfs_flags_to_ioctl(ip->flags);

	if (copy_to_user(arg, &flags, sizeof(flags)))
		return -EFAULT;
	return 0;
}

169 170 171 172 173
static int check_flags(unsigned int flags)
{
	if (flags & ~(FS_IMMUTABLE_FL | FS_APPEND_FL | \
		      FS_NOATIME_FL | FS_NODUMP_FL | \
		      FS_SYNC_FL | FS_DIRSYNC_FL | \
L
Li Zefan 已提交
174 175
		      FS_NOCOMP_FL | FS_COMPR_FL |
		      FS_NOCOW_FL))
176 177 178 179 180 181 182 183
		return -EOPNOTSUPP;

	if ((flags & FS_NOCOMP_FL) && (flags & FS_COMPR_FL))
		return -EINVAL;

	return 0;
}

184 185
static int btrfs_ioctl_setflags(struct file *file, void __user *arg)
{
A
Al Viro 已提交
186
	struct inode *inode = file_inode(file);
187
	struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb);
188 189 190 191 192
	struct btrfs_inode *ip = BTRFS_I(inode);
	struct btrfs_root *root = ip->root;
	struct btrfs_trans_handle *trans;
	unsigned int flags, oldflags;
	int ret;
193 194
	u64 ip_oldflags;
	unsigned int i_oldflags;
195
	umode_t mode;
196

197 198 199
	if (!inode_owner_or_capable(inode))
		return -EPERM;

L
Li Zefan 已提交
200 201 202
	if (btrfs_root_readonly(root))
		return -EROFS;

203 204 205
	if (copy_from_user(&flags, arg, sizeof(flags)))
		return -EFAULT;

206 207 208
	ret = check_flags(flags);
	if (ret)
		return ret;
C
Christoph Hellwig 已提交
209

210 211 212 213
	ret = mnt_want_write_file(file);
	if (ret)
		return ret;

A
Al Viro 已提交
214
	inode_lock(inode);
215

216 217
	ip_oldflags = ip->flags;
	i_oldflags = inode->i_flags;
218
	mode = inode->i_mode;
219

220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252
	flags = btrfs_mask_flags(inode->i_mode, flags);
	oldflags = btrfs_flags_to_ioctl(ip->flags);
	if ((flags ^ oldflags) & (FS_APPEND_FL | FS_IMMUTABLE_FL)) {
		if (!capable(CAP_LINUX_IMMUTABLE)) {
			ret = -EPERM;
			goto out_unlock;
		}
	}

	if (flags & FS_SYNC_FL)
		ip->flags |= BTRFS_INODE_SYNC;
	else
		ip->flags &= ~BTRFS_INODE_SYNC;
	if (flags & FS_IMMUTABLE_FL)
		ip->flags |= BTRFS_INODE_IMMUTABLE;
	else
		ip->flags &= ~BTRFS_INODE_IMMUTABLE;
	if (flags & FS_APPEND_FL)
		ip->flags |= BTRFS_INODE_APPEND;
	else
		ip->flags &= ~BTRFS_INODE_APPEND;
	if (flags & FS_NODUMP_FL)
		ip->flags |= BTRFS_INODE_NODUMP;
	else
		ip->flags &= ~BTRFS_INODE_NODUMP;
	if (flags & FS_NOATIME_FL)
		ip->flags |= BTRFS_INODE_NOATIME;
	else
		ip->flags &= ~BTRFS_INODE_NOATIME;
	if (flags & FS_DIRSYNC_FL)
		ip->flags |= BTRFS_INODE_DIRSYNC;
	else
		ip->flags &= ~BTRFS_INODE_DIRSYNC;
253 254 255 256 257 258 259 260 261 262 263 264 265 266 267
	if (flags & FS_NOCOW_FL) {
		if (S_ISREG(mode)) {
			/*
			 * It's safe to turn csums off here, no extents exist.
			 * Otherwise we want the flag to reflect the real COW
			 * status of the file and will not set it.
			 */
			if (inode->i_size == 0)
				ip->flags |= BTRFS_INODE_NODATACOW
					   | BTRFS_INODE_NODATASUM;
		} else {
			ip->flags |= BTRFS_INODE_NODATACOW;
		}
	} else {
		/*
268
		 * Revert back under same assumptions as above
269 270 271 272 273 274 275 276 277
		 */
		if (S_ISREG(mode)) {
			if (inode->i_size == 0)
				ip->flags &= ~(BTRFS_INODE_NODATACOW
				             | BTRFS_INODE_NODATASUM);
		} else {
			ip->flags &= ~BTRFS_INODE_NODATACOW;
		}
	}
278

279 280 281 282 283 284 285 286
	/*
	 * The COMPRESS flag can only be changed by users, while the NOCOMPRESS
	 * flag may be changed automatically if compression code won't make
	 * things smaller.
	 */
	if (flags & FS_NOCOMP_FL) {
		ip->flags &= ~BTRFS_INODE_COMPRESS;
		ip->flags |= BTRFS_INODE_NOCOMPRESS;
287 288 289 290

		ret = btrfs_set_prop(inode, "btrfs.compression", NULL, 0, 0);
		if (ret && ret != -ENODATA)
			goto out_drop;
291
	} else if (flags & FS_COMPR_FL) {
292 293
		const char *comp;

294 295
		ip->flags |= BTRFS_INODE_COMPRESS;
		ip->flags &= ~BTRFS_INODE_NOCOMPRESS;
296

297
		if (fs_info->compress_type == BTRFS_COMPRESS_LZO)
298
			comp = "lzo";
N
Nick Terrell 已提交
299
		else if (fs_info->compress_type == BTRFS_COMPRESS_ZLIB)
300
			comp = "zlib";
N
Nick Terrell 已提交
301 302
		else
			comp = "zstd";
303 304 305 306 307
		ret = btrfs_set_prop(inode, "btrfs.compression",
				     comp, strlen(comp), 0);
		if (ret)
			goto out_drop;

L
Li Zefan 已提交
308
	} else {
309 310 311
		ret = btrfs_set_prop(inode, "btrfs.compression", NULL, 0, 0);
		if (ret && ret != -ENODATA)
			goto out_drop;
L
Li Zefan 已提交
312
		ip->flags &= ~(BTRFS_INODE_COMPRESS | BTRFS_INODE_NOCOMPRESS);
313
	}
314

315
	trans = btrfs_start_transaction(root, 1);
316 317 318 319
	if (IS_ERR(trans)) {
		ret = PTR_ERR(trans);
		goto out_drop;
	}
320

321
	btrfs_update_iflags(inode);
322
	inode_inc_iversion(inode);
323
	inode->i_ctime = current_time(inode);
324 325
	ret = btrfs_update_inode(trans, root, inode);

326
	btrfs_end_transaction(trans);
327 328 329 330 331
 out_drop:
	if (ret) {
		ip->flags = ip_oldflags;
		inode->i_flags = i_oldflags;
	}
332 333

 out_unlock:
A
Al Viro 已提交
334
	inode_unlock(inode);
335
	mnt_drop_write_file(file);
336
	return ret;
337 338 339 340
}

static int btrfs_ioctl_getversion(struct file *file, int __user *arg)
{
A
Al Viro 已提交
341
	struct inode *inode = file_inode(file);
342 343 344

	return put_user(inode->i_generation, arg);
}
C
Christoph Hellwig 已提交
345

346 347
static noinline int btrfs_ioctl_fitrim(struct file *file, void __user *arg)
{
348 349
	struct inode *inode = file_inode(file);
	struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb);
350 351 352 353 354
	struct btrfs_device *device;
	struct request_queue *q;
	struct fstrim_range range;
	u64 minlen = ULLONG_MAX;
	u64 num_devices = 0;
355
	u64 total_bytes = btrfs_super_total_bytes(fs_info->super_copy);
356 357 358 359 360
	int ret;

	if (!capable(CAP_SYS_ADMIN))
		return -EPERM;

361 362 363
	rcu_read_lock();
	list_for_each_entry_rcu(device, &fs_info->fs_devices->devices,
				dev_list) {
364 365 366 367 368
		if (!device->bdev)
			continue;
		q = bdev_get_queue(device->bdev);
		if (blk_queue_discard(q)) {
			num_devices++;
369
			minlen = min_t(u64, q->limits.discard_granularity,
370 371 372
				     minlen);
		}
	}
373
	rcu_read_unlock();
374

375 376 377 378
	if (!num_devices)
		return -EOPNOTSUPP;
	if (copy_from_user(&range, arg, sizeof(range)))
		return -EFAULT;
379 380
	if (range.start > total_bytes ||
	    range.len < fs_info->sb->s_blocksize)
381
		return -EINVAL;
382

383
	range.len = min(range.len, total_bytes - range.start);
384
	range.minlen = max(range.minlen, minlen);
385
	ret = btrfs_trim_fs(fs_info, &range);
386 387 388 389 390 391 392 393 394
	if (ret < 0)
		return ret;

	if (copy_to_user(arg, &range, sizeof(range)))
		return -EFAULT;

	return 0;
}

395 396
int btrfs_is_empty_uuid(u8 *uuid)
{
C
Chris Mason 已提交
397 398 399 400 401 402 403
	int i;

	for (i = 0; i < BTRFS_UUID_SIZE; i++) {
		if (uuid[i])
			return 0;
	}
	return 1;
404 405
}

406
static noinline int create_subvol(struct inode *dir,
407
				  struct dentry *dentry,
408
				  const char *name, int namelen,
A
Arne Jansen 已提交
409
				  u64 *async_transid,
410
				  struct btrfs_qgroup_inherit *inherit)
C
Christoph Hellwig 已提交
411
{
412
	struct btrfs_fs_info *fs_info = btrfs_sb(dir->i_sb);
C
Christoph Hellwig 已提交
413 414
	struct btrfs_trans_handle *trans;
	struct btrfs_key key;
415
	struct btrfs_root_item *root_item;
C
Christoph Hellwig 已提交
416 417
	struct btrfs_inode_item *inode_item;
	struct extent_buffer *leaf;
418
	struct btrfs_root *root = BTRFS_I(dir)->root;
419
	struct btrfs_root *new_root;
420
	struct btrfs_block_rsv block_rsv;
421
	struct timespec cur_time = current_time(dir);
422
	struct inode *inode;
C
Christoph Hellwig 已提交
423 424 425 426
	int ret;
	int err;
	u64 objectid;
	u64 new_dirid = BTRFS_FIRST_FREE_OBJECTID;
427
	u64 index = 0;
428
	u64 qgroup_reserved;
429
	uuid_le new_uuid;
C
Christoph Hellwig 已提交
430

431 432 433 434
	root_item = kzalloc(sizeof(*root_item), GFP_KERNEL);
	if (!root_item)
		return -ENOMEM;

435
	ret = btrfs_find_free_objectid(fs_info->tree_root, &objectid);
436
	if (ret)
437
		goto fail_free;
438

439 440
	/*
	 * Don't create subvolume whose level is not zero. Or qgroup will be
441
	 * screwed up since it assumes subvolume qgroup's level to be 0.
442
	 */
443 444 445 446
	if (btrfs_qgroup_level(objectid)) {
		ret = -ENOSPC;
		goto fail_free;
	}
447

448
	btrfs_init_block_rsv(&block_rsv, BTRFS_BLOCK_RSV_TEMP);
J
Josef Bacik 已提交
449
	/*
450 451
	 * The same as the snapshot creation, please see the comment
	 * of create_snapshot().
J
Josef Bacik 已提交
452
	 */
453
	ret = btrfs_subvolume_reserve_metadata(root, &block_rsv,
454
					       8, &qgroup_reserved, false);
455
	if (ret)
456
		goto fail_free;
457 458 459 460

	trans = btrfs_start_transaction(root, 0);
	if (IS_ERR(trans)) {
		ret = PTR_ERR(trans);
461
		btrfs_subvolume_release_metadata(fs_info, &block_rsv);
462
		goto fail_free;
463 464 465
	}
	trans->block_rsv = &block_rsv;
	trans->bytes_reserved = block_rsv.size;
C
Christoph Hellwig 已提交
466

467
	ret = btrfs_qgroup_inherit(trans, fs_info, 0, objectid, inherit);
A
Arne Jansen 已提交
468 469 470
	if (ret)
		goto fail;

471
	leaf = btrfs_alloc_tree_block(trans, root, 0, objectid, NULL, 0, 0, 0);
472 473 474 475
	if (IS_ERR(leaf)) {
		ret = PTR_ERR(leaf);
		goto fail;
	}
C
Christoph Hellwig 已提交
476

477
	memzero_extent_buffer(leaf, 0, sizeof(struct btrfs_header));
C
Christoph Hellwig 已提交
478 479
	btrfs_set_header_bytenr(leaf, leaf->start);
	btrfs_set_header_generation(leaf, trans->transid);
480
	btrfs_set_header_backref_rev(leaf, BTRFS_MIXED_BACKREF_REV);
C
Christoph Hellwig 已提交
481 482
	btrfs_set_header_owner(leaf, objectid);

483 484
	write_extent_buffer_fsid(leaf, fs_info->fsid);
	write_extent_buffer_chunk_tree_uuid(leaf, fs_info->chunk_tree_uuid);
C
Christoph Hellwig 已提交
485 486
	btrfs_mark_buffer_dirty(leaf);

487
	inode_item = &root_item->inode;
488 489 490
	btrfs_set_stack_inode_generation(inode_item, 1);
	btrfs_set_stack_inode_size(inode_item, 3);
	btrfs_set_stack_inode_nlink(inode_item, 1);
491
	btrfs_set_stack_inode_nbytes(inode_item,
492
				     fs_info->nodesize);
493
	btrfs_set_stack_inode_mode(inode_item, S_IFDIR | 0755);
C
Christoph Hellwig 已提交
494

495 496
	btrfs_set_root_flags(root_item, 0);
	btrfs_set_root_limit(root_item, 0);
497
	btrfs_set_stack_inode_flags(inode_item, BTRFS_INODE_ROOT_ITEM_INIT);
498

499 500 501 502 503 504
	btrfs_set_root_bytenr(root_item, leaf->start);
	btrfs_set_root_generation(root_item, trans->transid);
	btrfs_set_root_level(root_item, 0);
	btrfs_set_root_refs(root_item, 1);
	btrfs_set_root_used(root_item, leaf->len);
	btrfs_set_root_last_snapshot(root_item, 0);
C
Christoph Hellwig 已提交
505

506 507
	btrfs_set_root_generation_v2(root_item,
			btrfs_root_generation(root_item));
508
	uuid_le_gen(&new_uuid);
509 510 511 512 513 514
	memcpy(root_item->uuid, new_uuid.b, BTRFS_UUID_SIZE);
	btrfs_set_stack_timespec_sec(&root_item->otime, cur_time.tv_sec);
	btrfs_set_stack_timespec_nsec(&root_item->otime, cur_time.tv_nsec);
	root_item->ctime = root_item->otime;
	btrfs_set_root_ctransid(root_item, trans->transid);
	btrfs_set_root_otransid(root_item, trans->transid);
C
Christoph Hellwig 已提交
515

516
	btrfs_tree_unlock(leaf);
C
Christoph Hellwig 已提交
517 518 519
	free_extent_buffer(leaf);
	leaf = NULL;

520
	btrfs_set_root_dirid(root_item, new_dirid);
C
Christoph Hellwig 已提交
521 522

	key.objectid = objectid;
523
	key.offset = 0;
524
	key.type = BTRFS_ROOT_ITEM_KEY;
525
	ret = btrfs_insert_root(trans, fs_info->tree_root, &key,
526
				root_item);
C
Christoph Hellwig 已提交
527 528 529
	if (ret)
		goto fail;

530
	key.offset = (u64)-1;
531
	new_root = btrfs_read_fs_root_no_name(fs_info, &key);
532 533
	if (IS_ERR(new_root)) {
		ret = PTR_ERR(new_root);
534
		btrfs_abort_transaction(trans, ret);
535 536
		goto fail;
	}
537 538 539

	btrfs_record_root_in_trans(trans, new_root);

540
	ret = btrfs_create_subvol_root(trans, new_root, root, new_dirid);
541 542
	if (ret) {
		/* We potentially lose an unused inode item here */
543
		btrfs_abort_transaction(trans, ret);
544 545 546
		goto fail;
	}

547 548 549 550
	mutex_lock(&new_root->objectid_mutex);
	new_root->highest_objectid = new_dirid;
	mutex_unlock(&new_root->objectid_mutex);

C
Christoph Hellwig 已提交
551 552 553
	/*
	 * insert the directory item
	 */
554
	ret = btrfs_set_inode_index(BTRFS_I(dir), &index);
555
	if (ret) {
556
		btrfs_abort_transaction(trans, ret);
557 558
		goto fail;
	}
559 560

	ret = btrfs_insert_dir_item(trans, root,
561
				    name, namelen, BTRFS_I(dir), &key,
562
				    BTRFS_FT_DIR, index);
563
	if (ret) {
564
		btrfs_abort_transaction(trans, ret);
C
Christoph Hellwig 已提交
565
		goto fail;
566
	}
567

568
	btrfs_i_size_write(BTRFS_I(dir), dir->i_size + namelen * 2);
569 570 571
	ret = btrfs_update_inode(trans, root, dir);
	BUG_ON(ret);

572
	ret = btrfs_add_root_ref(trans, fs_info,
573
				 objectid, root->root_key.objectid,
574
				 btrfs_ino(BTRFS_I(dir)), index, name, namelen);
575
	BUG_ON(ret);
C
Christoph Hellwig 已提交
576

577
	ret = btrfs_uuid_tree_add(trans, fs_info, root_item->uuid,
578
				  BTRFS_UUID_KEY_SUBVOL, objectid);
579
	if (ret)
580
		btrfs_abort_transaction(trans, ret);
581

C
Christoph Hellwig 已提交
582
fail:
583
	kfree(root_item);
584 585
	trans->block_rsv = NULL;
	trans->bytes_reserved = 0;
586
	btrfs_subvolume_release_metadata(fs_info, &block_rsv);
587

S
Sage Weil 已提交
588 589
	if (async_transid) {
		*async_transid = trans->transid;
590
		err = btrfs_commit_transaction_async(trans, 1);
591
		if (err)
592
			err = btrfs_commit_transaction(trans);
S
Sage Weil 已提交
593
	} else {
594
		err = btrfs_commit_transaction(trans);
S
Sage Weil 已提交
595
	}
C
Christoph Hellwig 已提交
596 597
	if (err && !ret)
		ret = err;
598

599 600
	if (!ret) {
		inode = btrfs_lookup_dentry(dir, dentry);
601 602
		if (IS_ERR(inode))
			return PTR_ERR(inode);
603 604
		d_instantiate(dentry, inode);
	}
C
Christoph Hellwig 已提交
605
	return ret;
606 607 608 609

fail_free:
	kfree(root_item);
	return ret;
C
Christoph Hellwig 已提交
610 611
}

612
static int create_snapshot(struct btrfs_root *root, struct inode *dir,
613
			   struct dentry *dentry,
614 615
			   u64 *async_transid, bool readonly,
			   struct btrfs_qgroup_inherit *inherit)
C
Christoph Hellwig 已提交
616
{
617
	struct btrfs_fs_info *fs_info = btrfs_sb(dir->i_sb);
618
	struct inode *inode;
C
Christoph Hellwig 已提交
619 620
	struct btrfs_pending_snapshot *pending_snapshot;
	struct btrfs_trans_handle *trans;
621
	int ret;
C
Christoph Hellwig 已提交
622

623
	if (!test_bit(BTRFS_ROOT_REF_COWS, &root->state))
C
Christoph Hellwig 已提交
624 625
		return -EINVAL;

626
	pending_snapshot = kzalloc(sizeof(*pending_snapshot), GFP_KERNEL);
627 628 629
	if (!pending_snapshot)
		return -ENOMEM;

630
	pending_snapshot->root_item = kzalloc(sizeof(struct btrfs_root_item),
631
			GFP_KERNEL);
632 633
	pending_snapshot->path = btrfs_alloc_path();
	if (!pending_snapshot->root_item || !pending_snapshot->path) {
634 635 636 637
		ret = -ENOMEM;
		goto free_pending;
	}

638
	atomic_inc(&root->will_be_snapshotted);
639
	smp_mb__after_atomic();
640 641 642
	/* wait for no snapshot writes */
	wait_event(root->subv_writers->wait,
		   percpu_counter_sum(&root->subv_writers->counter) == 0);
643

644 645
	ret = btrfs_start_delalloc_inodes(root, 0);
	if (ret)
646
		goto dec_and_free;
647

648
	btrfs_wait_ordered_extents(root, U64_MAX, 0, (u64)-1);
649

650 651
	btrfs_init_block_rsv(&pending_snapshot->block_rsv,
			     BTRFS_BLOCK_RSV_TEMP);
652 653 654 655 656 657
	/*
	 * 1 - parent dir inode
	 * 2 - dir entries
	 * 1 - root item
	 * 2 - root ref/backref
	 * 1 - root of snapshot
658
	 * 1 - UUID item
659 660
	 */
	ret = btrfs_subvolume_reserve_metadata(BTRFS_I(dir)->root,
661
					&pending_snapshot->block_rsv, 8,
662 663
					&pending_snapshot->qgroup_reserved,
					false);
664
	if (ret)
665
		goto dec_and_free;
666

667
	pending_snapshot->dentry = dentry;
C
Christoph Hellwig 已提交
668
	pending_snapshot->root = root;
L
Li Zefan 已提交
669
	pending_snapshot->readonly = readonly;
670
	pending_snapshot->dir = dir;
671
	pending_snapshot->inherit = inherit;
672

673
	trans = btrfs_start_transaction(root, 0);
674 675 676 677 678
	if (IS_ERR(trans)) {
		ret = PTR_ERR(trans);
		goto fail;
	}

679
	spin_lock(&fs_info->trans_lock);
C
Christoph Hellwig 已提交
680 681
	list_add(&pending_snapshot->list,
		 &trans->transaction->pending_snapshots);
682
	spin_unlock(&fs_info->trans_lock);
S
Sage Weil 已提交
683 684
	if (async_transid) {
		*async_transid = trans->transid;
685
		ret = btrfs_commit_transaction_async(trans, 1);
686
		if (ret)
687
			ret = btrfs_commit_transaction(trans);
S
Sage Weil 已提交
688
	} else {
689
		ret = btrfs_commit_transaction(trans);
S
Sage Weil 已提交
690
	}
691
	if (ret)
692
		goto fail;
693 694 695 696 697

	ret = pending_snapshot->error;
	if (ret)
		goto fail;

698 699 700 701
	ret = btrfs_orphan_cleanup(pending_snapshot->snap);
	if (ret)
		goto fail;

702
	inode = btrfs_lookup_dentry(d_inode(dentry->d_parent), dentry);
703 704 705 706
	if (IS_ERR(inode)) {
		ret = PTR_ERR(inode);
		goto fail;
	}
707

708 709 710
	d_instantiate(dentry, inode);
	ret = 0;
fail:
711
	btrfs_subvolume_release_metadata(fs_info, &pending_snapshot->block_rsv);
712
dec_and_free:
713 714
	if (atomic_dec_and_test(&root->will_be_snapshotted))
		wake_up_atomic_t(&root->will_be_snapshotted);
715 716
free_pending:
	kfree(pending_snapshot->root_item);
717
	btrfs_free_path(pending_snapshot->path);
718 719
	kfree(pending_snapshot);

C
Christoph Hellwig 已提交
720 721 722
	return ret;
}

723 724 725 726 727 728 729 730 731 732 733
/*  copy of may_delete in fs/namei.c()
 *	Check whether we can remove a link victim from directory dir, check
 *  whether the type of victim is right.
 *  1. We can't do it if dir is read-only (done in permission())
 *  2. We should have write and exec permissions on dir
 *  3. We can't remove anything from append-only dir
 *  4. We can't do anything with immutable dir (done in permission())
 *  5. If the sticky bit on dir is set we should either
 *	a. be owner of dir, or
 *	b. be owner of victim, or
 *	c. have CAP_FOWNER capability
734
 *  6. If the victim is append-only or immutable we can't do anything with
735 736 737 738 739 740 741 742
 *     links pointing to it.
 *  7. If we were asked to remove a directory and victim isn't one - ENOTDIR.
 *  8. If we were asked to remove a non-directory and victim isn't one - EISDIR.
 *  9. We can't remove a root or mountpoint.
 * 10. We don't allow removal of NFS sillyrenamed files; it's handled by
 *     nfs_async_unlink().
 */

743
static int btrfs_may_delete(struct inode *dir, struct dentry *victim, int isdir)
744 745 746
{
	int error;

747
	if (d_really_is_negative(victim))
748 749
		return -ENOENT;

750
	BUG_ON(d_inode(victim->d_parent) != dir);
751
	audit_inode_child(dir, victim, AUDIT_TYPE_CHILD_DELETE);
752 753 754 755 756 757

	error = inode_permission(dir, MAY_WRITE | MAY_EXEC);
	if (error)
		return error;
	if (IS_APPEND(dir))
		return -EPERM;
758 759
	if (check_sticky(dir, d_inode(victim)) || IS_APPEND(d_inode(victim)) ||
	    IS_IMMUTABLE(d_inode(victim)) || IS_SWAPFILE(d_inode(victim)))
760 761
		return -EPERM;
	if (isdir) {
762
		if (!d_is_dir(victim))
763 764 765
			return -ENOTDIR;
		if (IS_ROOT(victim))
			return -EBUSY;
766
	} else if (d_is_dir(victim))
767 768 769 770 771 772 773 774
		return -EISDIR;
	if (IS_DEADDIR(dir))
		return -ENOENT;
	if (victim->d_flags & DCACHE_NFSFS_RENAMED)
		return -EBUSY;
	return 0;
}

775 776 777
/* copy of may_create in fs/namei.c() */
static inline int btrfs_may_create(struct inode *dir, struct dentry *child)
{
778
	if (d_really_is_positive(child))
779 780 781 782 783 784 785 786 787 788 789
		return -EEXIST;
	if (IS_DEADDIR(dir))
		return -ENOENT;
	return inode_permission(dir, MAY_WRITE | MAY_EXEC);
}

/*
 * Create a new subvolume below @parent.  This is largely modeled after
 * sys_mkdirat and vfs_mkdir, but we only do a single component lookup
 * inside this filesystem so it's quite a bit simpler.
 */
A
Al Viro 已提交
790
static noinline int btrfs_mksubvol(const struct path *parent,
791
				   const char *name, int namelen,
S
Sage Weil 已提交
792
				   struct btrfs_root *snap_src,
A
Arne Jansen 已提交
793
				   u64 *async_transid, bool readonly,
794
				   struct btrfs_qgroup_inherit *inherit)
795
{
796 797
	struct inode *dir = d_inode(parent->dentry);
	struct btrfs_fs_info *fs_info = btrfs_sb(dir->i_sb);
798 799 800
	struct dentry *dentry;
	int error;

801 802 803
	error = down_write_killable_nested(&dir->i_rwsem, I_MUTEX_PARENT);
	if (error == -EINTR)
		return error;
804 805 806 807 808 809

	dentry = lookup_one_len(name, parent->dentry, namelen);
	error = PTR_ERR(dentry);
	if (IS_ERR(dentry))
		goto out_unlock;

810
	error = btrfs_may_create(dir, dentry);
811
	if (error)
812
		goto out_dput;
813

C
Chris Mason 已提交
814 815 816 817 818 819 820 821 822 823
	/*
	 * even if this name doesn't exist, we may get hash collisions.
	 * check for them now when we can safely fail
	 */
	error = btrfs_check_dir_item_collision(BTRFS_I(dir)->root,
					       dir->i_ino, name,
					       namelen);
	if (error)
		goto out_dput;

824
	down_read(&fs_info->subvol_sem);
825 826 827 828

	if (btrfs_root_refs(&BTRFS_I(dir)->root->root_item) == 0)
		goto out_up_read;

829
	if (snap_src) {
830
		error = create_snapshot(snap_src, dir, dentry,
A
Arne Jansen 已提交
831
					async_transid, readonly, inherit);
832
	} else {
833 834
		error = create_subvol(dir, dentry, name, namelen,
				      async_transid, inherit);
835
	}
836 837 838
	if (!error)
		fsnotify_mkdir(dir, dentry);
out_up_read:
839
	up_read(&fs_info->subvol_sem);
840 841 842
out_dput:
	dput(dentry);
out_unlock:
A
Al Viro 已提交
843
	inode_unlock(dir);
844 845 846
	return error;
}

C
Chris Mason 已提交
847 848 849 850 851 852 853
/*
 * When we're defragging a range, we don't want to kick it off again
 * if it is really just waiting for delalloc to send it down.
 * If we find a nice big extent or delalloc range for the bytes in the
 * file you want to defrag, we return 0 to let you know to skip this
 * part of the file
 */
854
static int check_defrag_in_cache(struct inode *inode, u64 offset, u32 thresh)
C
Chris Mason 已提交
855 856 857 858 859 860 861
{
	struct extent_io_tree *io_tree = &BTRFS_I(inode)->io_tree;
	struct extent_map *em = NULL;
	struct extent_map_tree *em_tree = &BTRFS_I(inode)->extent_tree;
	u64 end;

	read_lock(&em_tree->lock);
862
	em = lookup_extent_mapping(em_tree, offset, PAGE_SIZE);
C
Chris Mason 已提交
863 864 865 866 867 868 869 870 871 872 873 874 875 876 877 878 879 880 881 882 883 884 885 886 887 888
	read_unlock(&em_tree->lock);

	if (em) {
		end = extent_map_end(em);
		free_extent_map(em);
		if (end - offset > thresh)
			return 0;
	}
	/* if we already have a nice delalloc here, just stop */
	thresh /= 2;
	end = count_range_bits(io_tree, &offset, offset + thresh,
			       thresh, EXTENT_DELALLOC, 1);
	if (end >= thresh)
		return 0;
	return 1;
}

/*
 * helper function to walk through a file and find extents
 * newer than a specific transid, and smaller than thresh.
 *
 * This is used by the defragging code to find new and small
 * extents
 */
static int find_new_extents(struct btrfs_root *root,
			    struct inode *inode, u64 newer_than,
889
			    u64 *off, u32 thresh)
C
Chris Mason 已提交
890 891 892 893 894 895 896
{
	struct btrfs_path *path;
	struct btrfs_key min_key;
	struct extent_buffer *leaf;
	struct btrfs_file_extent_item *extent;
	int type;
	int ret;
897
	u64 ino = btrfs_ino(BTRFS_I(inode));
C
Chris Mason 已提交
898 899 900 901 902

	path = btrfs_alloc_path();
	if (!path)
		return -ENOMEM;

903
	min_key.objectid = ino;
C
Chris Mason 已提交
904 905 906
	min_key.type = BTRFS_EXTENT_DATA_KEY;
	min_key.offset = *off;

907
	while (1) {
908
		ret = btrfs_search_forward(root, &min_key, path, newer_than);
C
Chris Mason 已提交
909 910
		if (ret != 0)
			goto none;
911
process_slot:
912
		if (min_key.objectid != ino)
C
Chris Mason 已提交
913 914 915 916 917 918 919 920 921 922 923 924 925 926 927 928 929
			goto none;
		if (min_key.type != BTRFS_EXTENT_DATA_KEY)
			goto none;

		leaf = path->nodes[0];
		extent = btrfs_item_ptr(leaf, path->slots[0],
					struct btrfs_file_extent_item);

		type = btrfs_file_extent_type(leaf, extent);
		if (type == BTRFS_FILE_EXTENT_REG &&
		    btrfs_file_extent_num_bytes(leaf, extent) < thresh &&
		    check_defrag_in_cache(inode, min_key.offset, thresh)) {
			*off = min_key.offset;
			btrfs_free_path(path);
			return 0;
		}

930 931 932 933 934 935
		path->slots[0]++;
		if (path->slots[0] < btrfs_header_nritems(leaf)) {
			btrfs_item_key_to_cpu(leaf, &min_key, path->slots[0]);
			goto process_slot;
		}

C
Chris Mason 已提交
936 937 938 939 940 941 942 943 944 945 946
		if (min_key.offset == (u64)-1)
			goto none;

		min_key.offset++;
		btrfs_release_path(path);
	}
none:
	btrfs_free_path(path);
	return -ENOENT;
}

L
Li Zefan 已提交
947
static struct extent_map *defrag_lookup_extent(struct inode *inode, u64 start)
948 949
{
	struct extent_map_tree *em_tree = &BTRFS_I(inode)->extent_tree;
L
Li Zefan 已提交
950 951
	struct extent_io_tree *io_tree = &BTRFS_I(inode)->io_tree;
	struct extent_map *em;
952
	u64 len = PAGE_SIZE;
953

L
Li Zefan 已提交
954 955 956 957
	/*
	 * hopefully we have this extent in the tree already, try without
	 * the full extent lock
	 */
958
	read_lock(&em_tree->lock);
L
Li Zefan 已提交
959
	em = lookup_extent_mapping(em_tree, start, len);
960 961
	read_unlock(&em_tree->lock);

L
Li Zefan 已提交
962
	if (!em) {
963 964 965
		struct extent_state *cached = NULL;
		u64 end = start + len - 1;

L
Li Zefan 已提交
966
		/* get the big lock and read metadata off disk */
967
		lock_extent_bits(io_tree, start, end, &cached);
968
		em = btrfs_get_extent(BTRFS_I(inode), NULL, 0, start, len, 0);
969
		unlock_extent_cached(io_tree, start, end, &cached, GFP_NOFS);
L
Li Zefan 已提交
970 971 972 973 974 975 976

		if (IS_ERR(em))
			return NULL;
	}

	return em;
}
977

L
Li Zefan 已提交
978 979 980 981 982 983 984 985 986 987
static bool defrag_check_next_extent(struct inode *inode, struct extent_map *em)
{
	struct extent_map *next;
	bool ret = true;

	/* this is the last extent */
	if (em->start + em->len >= i_size_read(inode))
		return false;

	next = defrag_lookup_extent(inode, em->start + em->len);
988 989 990
	if (!next || next->block_start >= EXTENT_MAP_LAST_BYTE)
		ret = false;
	else if ((em->block_start + em->block_len == next->block_start) &&
991
		 (em->block_len > SZ_128K && next->block_len > SZ_128K))
L
Li Zefan 已提交
992 993 994
		ret = false;

	free_extent_map(next);
995 996 997
	return ret;
}

998
static int should_defrag_range(struct inode *inode, u64 start, u32 thresh,
999 1000
			       u64 *last_len, u64 *skip, u64 *defrag_end,
			       int compress)
1001
{
L
Li Zefan 已提交
1002
	struct extent_map *em;
1003
	int ret = 1;
L
Li Zefan 已提交
1004
	bool next_mergeable = true;
1005
	bool prev_mergeable = true;
1006 1007

	/*
1008
	 * make sure that once we start defragging an extent, we keep on
1009 1010 1011 1012 1013 1014 1015
	 * defragging it
	 */
	if (start < *defrag_end)
		return 1;

	*skip = 0;

L
Li Zefan 已提交
1016 1017 1018
	em = defrag_lookup_extent(inode, start);
	if (!em)
		return 0;
1019 1020

	/* this will cover holes, and inline extents */
1021
	if (em->block_start >= EXTENT_MAP_LAST_BYTE) {
1022
		ret = 0;
1023 1024 1025
		goto out;
	}

1026 1027 1028
	if (!*defrag_end)
		prev_mergeable = false;

L
Li Zefan 已提交
1029
	next_mergeable = defrag_check_next_extent(inode, em);
1030
	/*
L
Li Zefan 已提交
1031 1032
	 * we hit a real extent, if it is big or the next extent is not a
	 * real extent, don't bother defragging it
1033
	 */
1034
	if (!compress && (*last_len == 0 || *last_len >= thresh) &&
1035
	    (em->len >= thresh || (!next_mergeable && !prev_mergeable)))
1036
		ret = 0;
1037
out:
1038 1039 1040 1041 1042 1043 1044 1045 1046 1047 1048 1049 1050 1051 1052 1053 1054 1055 1056 1057
	/*
	 * last_len ends up being a counter of how many bytes we've defragged.
	 * every time we choose not to defrag an extent, we reset *last_len
	 * so that the next tiny extent will force a defrag.
	 *
	 * The end result of this is that tiny extents before a single big
	 * extent will force at least part of that big extent to be defragged.
	 */
	if (ret) {
		*defrag_end = extent_map_end(em);
	} else {
		*last_len = 0;
		*skip = extent_map_end(em);
		*defrag_end = 0;
	}

	free_extent_map(em);
	return ret;
}

C
Chris Mason 已提交
1058 1059 1060 1061 1062 1063 1064 1065 1066 1067 1068 1069 1070 1071 1072
/*
 * it doesn't do much good to defrag one or two pages
 * at a time.  This pulls in a nice chunk of pages
 * to COW and defrag.
 *
 * It also makes sure the delalloc code has enough
 * dirty data to avoid making new small extents as part
 * of the defrag
 *
 * It's a good idea to start RA on this range
 * before calling this.
 */
static int cluster_pages_for_defrag(struct inode *inode,
				    struct page **pages,
				    unsigned long start_index,
1073
				    unsigned long num_pages)
C
Christoph Hellwig 已提交
1074
{
C
Chris Mason 已提交
1075 1076 1077 1078
	unsigned long file_end;
	u64 isize = i_size_read(inode);
	u64 page_start;
	u64 page_end;
1079
	u64 page_cnt;
C
Chris Mason 已提交
1080 1081 1082
	int ret;
	int i;
	int i_done;
1083
	struct btrfs_ordered_extent *ordered;
C
Chris Mason 已提交
1084
	struct extent_state *cached_state = NULL;
1085
	struct extent_io_tree *tree;
1086
	struct extent_changeset *data_reserved = NULL;
1087
	gfp_t mask = btrfs_alloc_write_mask(inode->i_mapping);
C
Chris Mason 已提交
1088

1089
	file_end = (isize - 1) >> PAGE_SHIFT;
1090 1091 1092 1093
	if (!isize || start_index > file_end)
		return 0;

	page_cnt = min_t(u64, (u64)num_pages, (u64)file_end - start_index + 1);
C
Chris Mason 已提交
1094

1095
	ret = btrfs_delalloc_reserve_space(inode, &data_reserved,
1096 1097
			start_index << PAGE_SHIFT,
			page_cnt << PAGE_SHIFT);
C
Chris Mason 已提交
1098 1099 1100
	if (ret)
		return ret;
	i_done = 0;
1101
	tree = &BTRFS_I(inode)->io_tree;
C
Chris Mason 已提交
1102 1103

	/* step one, lock all the pages */
1104
	for (i = 0; i < page_cnt; i++) {
C
Chris Mason 已提交
1105
		struct page *page;
1106
again:
1107
		page = find_or_create_page(inode->i_mapping,
1108
					   start_index + i, mask);
C
Chris Mason 已提交
1109 1110 1111
		if (!page)
			break;

1112
		page_start = page_offset(page);
1113
		page_end = page_start + PAGE_SIZE - 1;
1114
		while (1) {
1115
			lock_extent_bits(tree, page_start, page_end,
1116
					 &cached_state);
1117 1118
			ordered = btrfs_lookup_ordered_extent(inode,
							      page_start);
1119 1120
			unlock_extent_cached(tree, page_start, page_end,
					     &cached_state, GFP_NOFS);
1121 1122 1123 1124 1125 1126 1127
			if (!ordered)
				break;

			unlock_page(page);
			btrfs_start_ordered_extent(inode, ordered, 1);
			btrfs_put_ordered_extent(ordered);
			lock_page(page);
1128 1129 1130 1131 1132 1133
			/*
			 * we unlocked the page above, so we need check if
			 * it was released or not.
			 */
			if (page->mapping != inode->i_mapping) {
				unlock_page(page);
1134
				put_page(page);
1135 1136
				goto again;
			}
1137 1138
		}

C
Chris Mason 已提交
1139 1140 1141 1142 1143
		if (!PageUptodate(page)) {
			btrfs_readpage(NULL, page);
			lock_page(page);
			if (!PageUptodate(page)) {
				unlock_page(page);
1144
				put_page(page);
C
Chris Mason 已提交
1145 1146 1147 1148
				ret = -EIO;
				break;
			}
		}
1149 1150 1151

		if (page->mapping != inode->i_mapping) {
			unlock_page(page);
1152
			put_page(page);
1153 1154 1155
			goto again;
		}

C
Chris Mason 已提交
1156 1157 1158 1159 1160 1161 1162 1163 1164 1165 1166 1167 1168 1169 1170 1171 1172
		pages[i] = page;
		i_done++;
	}
	if (!i_done || ret)
		goto out;

	if (!(inode->i_sb->s_flags & MS_ACTIVE))
		goto out;

	/*
	 * so now we have a nice long stream of locked
	 * and up to date pages, lets wait on them
	 */
	for (i = 0; i < i_done; i++)
		wait_on_page_writeback(pages[i]);

	page_start = page_offset(pages[0]);
1173
	page_end = page_offset(pages[i_done - 1]) + PAGE_SIZE;
C
Chris Mason 已提交
1174 1175

	lock_extent_bits(&BTRFS_I(inode)->io_tree,
1176
			 page_start, page_end - 1, &cached_state);
C
Chris Mason 已提交
1177 1178
	clear_extent_bit(&BTRFS_I(inode)->io_tree, page_start,
			  page_end - 1, EXTENT_DIRTY | EXTENT_DELALLOC |
1179 1180
			  EXTENT_DO_ACCOUNTING | EXTENT_DEFRAG, 0, 0,
			  &cached_state, GFP_NOFS);
C
Chris Mason 已提交
1181

1182
	if (i_done != page_cnt) {
1183 1184 1185
		spin_lock(&BTRFS_I(inode)->lock);
		BTRFS_I(inode)->outstanding_extents++;
		spin_unlock(&BTRFS_I(inode)->lock);
1186
		btrfs_delalloc_release_space(inode, data_reserved,
1187 1188
				start_index << PAGE_SHIFT,
				(page_cnt - i_done) << PAGE_SHIFT);
C
Chris Mason 已提交
1189 1190 1191
	}


1192
	set_extent_defrag(&BTRFS_I(inode)->io_tree, page_start, page_end - 1,
1193
			  &cached_state);
C
Chris Mason 已提交
1194 1195 1196 1197 1198 1199 1200 1201 1202 1203 1204

	unlock_extent_cached(&BTRFS_I(inode)->io_tree,
			     page_start, page_end - 1, &cached_state,
			     GFP_NOFS);

	for (i = 0; i < i_done; i++) {
		clear_page_dirty_for_io(pages[i]);
		ClearPageChecked(pages[i]);
		set_page_extent_mapped(pages[i]);
		set_page_dirty(pages[i]);
		unlock_page(pages[i]);
1205
		put_page(pages[i]);
C
Chris Mason 已提交
1206
	}
1207
	extent_changeset_free(data_reserved);
C
Chris Mason 已提交
1208 1209 1210 1211
	return i_done;
out:
	for (i = 0; i < i_done; i++) {
		unlock_page(pages[i]);
1212
		put_page(pages[i]);
C
Chris Mason 已提交
1213
	}
1214
	btrfs_delalloc_release_space(inode, data_reserved,
1215 1216
			start_index << PAGE_SHIFT,
			page_cnt << PAGE_SHIFT);
1217
	extent_changeset_free(data_reserved);
C
Chris Mason 已提交
1218 1219 1220 1221 1222 1223 1224 1225
	return ret;

}

int btrfs_defrag_file(struct inode *inode, struct file *file,
		      struct btrfs_ioctl_defrag_range_args *range,
		      u64 newer_than, unsigned long max_to_defrag)
{
1226
	struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb);
C
Chris Mason 已提交
1227 1228
	struct btrfs_root *root = BTRFS_I(inode)->root;
	struct file_ra_state *ra = NULL;
C
Christoph Hellwig 已提交
1229
	unsigned long last_index;
1230
	u64 isize = i_size_read(inode);
1231 1232 1233
	u64 last_len = 0;
	u64 skip = 0;
	u64 defrag_end = 0;
C
Chris Mason 已提交
1234
	u64 newer_off = range->start;
C
Christoph Hellwig 已提交
1235
	unsigned long i;
1236
	unsigned long ra_index = 0;
C
Christoph Hellwig 已提交
1237
	int ret;
C
Chris Mason 已提交
1238
	int defrag_count = 0;
1239
	int compress_type = BTRFS_COMPRESS_ZLIB;
1240
	u32 extent_thresh = range->extent_thresh;
1241
	unsigned long max_cluster = SZ_256K >> PAGE_SHIFT;
1242
	unsigned long cluster = max_cluster;
1243
	u64 new_align = ~((u64)SZ_128K - 1);
C
Chris Mason 已提交
1244
	struct page **pages = NULL;
1245
	bool do_compress = range->flags & BTRFS_DEFRAG_RANGE_COMPRESS;
C
Chris Mason 已提交
1246

1247 1248 1249 1250 1251
	if (isize == 0)
		return 0;

	if (range->start >= isize)
		return -EINVAL;
1252

1253
	if (do_compress) {
1254 1255 1256 1257 1258
		if (range->compress_type > BTRFS_COMPRESS_TYPES)
			return -EINVAL;
		if (range->compress_type)
			compress_type = range->compress_type;
	}
C
Christoph Hellwig 已提交
1259

1260
	if (extent_thresh == 0)
1261
		extent_thresh = SZ_256K;
1262

C
Chris Mason 已提交
1263
	/*
1264 1265 1266
	 * If we were not given a file, allocate a readahead context. As
	 * readahead is just an optimization, defrag will work without it so
	 * we don't error out.
C
Chris Mason 已提交
1267 1268
	 */
	if (!file) {
1269
		ra = kzalloc(sizeof(*ra), GFP_KERNEL);
1270 1271
		if (ra)
			file_ra_state_init(ra, inode->i_mapping);
C
Chris Mason 已提交
1272 1273 1274 1275
	} else {
		ra = &file->f_ra;
	}

1276
	pages = kmalloc_array(max_cluster, sizeof(struct page *), GFP_KERNEL);
C
Chris Mason 已提交
1277 1278 1279 1280 1281 1282
	if (!pages) {
		ret = -ENOMEM;
		goto out_ra;
	}

	/* find the last page to defrag */
C
Chris Mason 已提交
1283
	if (range->start + range->len > range->start) {
1284
		last_index = min_t(u64, isize - 1,
1285
			 range->start + range->len - 1) >> PAGE_SHIFT;
C
Chris Mason 已提交
1286
	} else {
1287
		last_index = (isize - 1) >> PAGE_SHIFT;
C
Chris Mason 已提交
1288 1289
	}

C
Chris Mason 已提交
1290 1291
	if (newer_than) {
		ret = find_new_extents(root, inode, newer_than,
1292
				       &newer_off, SZ_64K);
C
Chris Mason 已提交
1293 1294 1295 1296 1297 1298
		if (!ret) {
			range->start = newer_off;
			/*
			 * we always align our defrag to help keep
			 * the extents in the file evenly spaced
			 */
1299
			i = (newer_off & new_align) >> PAGE_SHIFT;
C
Chris Mason 已提交
1300 1301 1302
		} else
			goto out_ra;
	} else {
1303
		i = range->start >> PAGE_SHIFT;
C
Chris Mason 已提交
1304 1305
	}
	if (!max_to_defrag)
1306
		max_to_defrag = last_index - i + 1;
C
Chris Mason 已提交
1307

L
Li Zefan 已提交
1308 1309 1310 1311 1312 1313 1314
	/*
	 * make writeback starts from i, so the defrag range can be
	 * written sequentially.
	 */
	if (i < inode->i_mapping->writeback_index)
		inode->i_mapping->writeback_index = i;

1315
	while (i <= last_index && defrag_count < max_to_defrag &&
1316
	       (i < DIV_ROUND_UP(i_size_read(inode), PAGE_SIZE))) {
C
Chris Mason 已提交
1317 1318 1319 1320 1321 1322 1323
		/*
		 * make sure we stop running if someone unmounts
		 * the FS
		 */
		if (!(inode->i_sb->s_flags & MS_ACTIVE))
			break;

1324 1325
		if (btrfs_defrag_cancelled(fs_info)) {
			btrfs_debug(fs_info, "defrag_file cancelled");
1326 1327 1328 1329
			ret = -EAGAIN;
			break;
		}

1330
		if (!should_defrag_range(inode, (u64)i << PAGE_SHIFT,
L
Li Zefan 已提交
1331
					 extent_thresh, &last_len, &skip,
1332
					 &defrag_end, do_compress)){
1333 1334 1335 1336 1337
			unsigned long next;
			/*
			 * the should_defrag function tells us how much to skip
			 * bump our counter by the suggested amount
			 */
1338
			next = DIV_ROUND_UP(skip, PAGE_SIZE);
1339 1340 1341
			i = max(i + 1, next);
			continue;
		}
1342 1343

		if (!newer_than) {
1344 1345
			cluster = (PAGE_ALIGN(defrag_end) >>
				   PAGE_SHIFT) - i;
1346 1347 1348 1349 1350 1351 1352
			cluster = min(cluster, max_cluster);
		} else {
			cluster = max_cluster;
		}

		if (i + cluster > ra_index) {
			ra_index = max(i, ra_index);
1353
			if (ra)
1354 1355
				page_cache_sync_readahead(inode->i_mapping, ra,
						file, ra_index, cluster);
1356
			ra_index += cluster;
1357
		}
1358

A
Al Viro 已提交
1359
		inode_lock(inode);
1360
		if (do_compress)
1361
			BTRFS_I(inode)->defrag_compress = compress_type;
1362
		ret = cluster_pages_for_defrag(inode, pages, i, cluster);
1363
		if (ret < 0) {
A
Al Viro 已提交
1364
			inode_unlock(inode);
C
Chris Mason 已提交
1365
			goto out_ra;
1366
		}
C
Chris Mason 已提交
1367 1368

		defrag_count += ret;
1369
		balance_dirty_pages_ratelimited(inode->i_mapping);
A
Al Viro 已提交
1370
		inode_unlock(inode);
C
Chris Mason 已提交
1371 1372 1373 1374 1375

		if (newer_than) {
			if (newer_off == (u64)-1)
				break;

1376 1377 1378
			if (ret > 0)
				i += ret;

C
Chris Mason 已提交
1379
			newer_off = max(newer_off + 1,
1380
					(u64)i << PAGE_SHIFT);
C
Chris Mason 已提交
1381

1382 1383
			ret = find_new_extents(root, inode, newer_than,
					       &newer_off, SZ_64K);
C
Chris Mason 已提交
1384 1385
			if (!ret) {
				range->start = newer_off;
1386
				i = (newer_off & new_align) >> PAGE_SHIFT;
C
Chris Mason 已提交
1387 1388
			} else {
				break;
C
Christoph Hellwig 已提交
1389
			}
C
Chris Mason 已提交
1390
		} else {
1391
			if (ret > 0) {
L
Li Zefan 已提交
1392
				i += ret;
1393
				last_len += ret << PAGE_SHIFT;
1394
			} else {
L
Li Zefan 已提交
1395
				i++;
1396 1397
				last_len = 0;
			}
C
Christoph Hellwig 已提交
1398 1399 1400
		}
	}

1401
	if ((range->flags & BTRFS_DEFRAG_RANGE_START_IO)) {
C
Chris Mason 已提交
1402
		filemap_flush(inode->i_mapping);
1403 1404 1405 1406
		if (test_bit(BTRFS_INODE_HAS_ASYNC_EXTENT,
			     &BTRFS_I(inode)->runtime_flags))
			filemap_flush(inode->i_mapping);
	}
C
Chris Mason 已提交
1407

1408
	if (do_compress) {
C
Chris Mason 已提交
1409 1410 1411 1412
		/* the filemap_flush will queue IO into the worker threads, but
		 * we have to make sure the IO is actually started and that
		 * ordered extents get created before we return
		 */
1413 1414 1415 1416 1417 1418
		atomic_inc(&fs_info->async_submit_draining);
		while (atomic_read(&fs_info->nr_async_submits) ||
		       atomic_read(&fs_info->async_delalloc_pages)) {
			wait_event(fs_info->async_submit_wait,
				   (atomic_read(&fs_info->nr_async_submits) == 0 &&
				    atomic_read(&fs_info->async_delalloc_pages) == 0));
C
Chris Mason 已提交
1419
		}
1420
		atomic_dec(&fs_info->async_submit_draining);
C
Chris Mason 已提交
1421 1422
	}

1423
	if (range->compress_type == BTRFS_COMPRESS_LZO) {
1424
		btrfs_set_fs_incompat(fs_info, COMPRESS_LZO);
N
Nick Terrell 已提交
1425 1426
	} else if (range->compress_type == BTRFS_COMPRESS_ZSTD) {
		btrfs_set_fs_incompat(fs_info, COMPRESS_ZSTD);
1427 1428
	}

1429
	ret = defrag_count;
1430

C
Chris Mason 已提交
1431
out_ra:
1432
	if (do_compress) {
A
Al Viro 已提交
1433
		inode_lock(inode);
1434
		BTRFS_I(inode)->defrag_compress = BTRFS_COMPRESS_NONE;
A
Al Viro 已提交
1435
		inode_unlock(inode);
1436
	}
C
Chris Mason 已提交
1437 1438 1439
	if (!file)
		kfree(ra);
	kfree(pages);
1440
	return ret;
C
Christoph Hellwig 已提交
1441 1442
}

1443
static noinline int btrfs_ioctl_resize(struct file *file,
1444
					void __user *arg)
C
Christoph Hellwig 已提交
1445
{
1446 1447
	struct inode *inode = file_inode(file);
	struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb);
C
Christoph Hellwig 已提交
1448 1449 1450
	u64 new_size;
	u64 old_size;
	u64 devid = 1;
1451
	struct btrfs_root *root = BTRFS_I(inode)->root;
C
Christoph Hellwig 已提交
1452 1453 1454 1455
	struct btrfs_ioctl_vol_args *vol_args;
	struct btrfs_trans_handle *trans;
	struct btrfs_device *device = NULL;
	char *sizestr;
1456
	char *retptr;
C
Christoph Hellwig 已提交
1457 1458 1459 1460
	char *devstr = NULL;
	int ret = 0;
	int mod = 0;

1461 1462 1463
	if (!capable(CAP_SYS_ADMIN))
		return -EPERM;

1464 1465 1466 1467
	ret = mnt_want_write_file(file);
	if (ret)
		return ret;

1468
	if (test_and_set_bit(BTRFS_FS_EXCL_OP, &fs_info->flags)) {
1469
		mnt_drop_write_file(file);
1470
		return BTRFS_ERROR_DEV_EXCL_RUN_IN_PROGRESS;
1471 1472
	}

1473
	mutex_lock(&fs_info->volume_mutex);
L
Li Zefan 已提交
1474
	vol_args = memdup_user(arg, sizeof(*vol_args));
1475 1476 1477 1478
	if (IS_ERR(vol_args)) {
		ret = PTR_ERR(vol_args);
		goto out;
	}
1479 1480

	vol_args->name[BTRFS_PATH_NAME_MAX] = '\0';
C
Christoph Hellwig 已提交
1481 1482 1483 1484 1485 1486 1487

	sizestr = vol_args->name;
	devstr = strchr(sizestr, ':');
	if (devstr) {
		sizestr = devstr + 1;
		*devstr = '\0';
		devstr = vol_args->name;
1488 1489 1490
		ret = kstrtoull(devstr, 10, &devid);
		if (ret)
			goto out_free;
1491 1492 1493 1494
		if (!devid) {
			ret = -EINVAL;
			goto out_free;
		}
1495
		btrfs_info(fs_info, "resizing devid %llu", devid);
C
Christoph Hellwig 已提交
1496
	}
M
Miao Xie 已提交
1497

1498
	device = btrfs_find_device(fs_info, devid, NULL, NULL);
C
Christoph Hellwig 已提交
1499
	if (!device) {
1500 1501
		btrfs_info(fs_info, "resizer unable to find device %llu",
			   devid);
1502
		ret = -ENODEV;
1503
		goto out_free;
C
Christoph Hellwig 已提交
1504
	}
M
Miao Xie 已提交
1505 1506

	if (!device->writeable) {
1507
		btrfs_info(fs_info,
1508
			   "resizer unable to apply on readonly device %llu",
1509
		       devid);
1510
		ret = -EPERM;
L
Liu Bo 已提交
1511 1512 1513
		goto out_free;
	}

C
Christoph Hellwig 已提交
1514 1515 1516 1517 1518 1519 1520 1521 1522 1523
	if (!strcmp(sizestr, "max"))
		new_size = device->bdev->bd_inode->i_size;
	else {
		if (sizestr[0] == '-') {
			mod = -1;
			sizestr++;
		} else if (sizestr[0] == '+') {
			mod = 1;
			sizestr++;
		}
1524 1525
		new_size = memparse(sizestr, &retptr);
		if (*retptr != '\0' || new_size == 0) {
C
Christoph Hellwig 已提交
1526
			ret = -EINVAL;
1527
			goto out_free;
C
Christoph Hellwig 已提交
1528 1529 1530
		}
	}

1531
	if (device->is_tgtdev_for_dev_replace) {
1532
		ret = -EPERM;
1533 1534 1535
		goto out_free;
	}

1536
	old_size = btrfs_device_get_total_bytes(device);
C
Christoph Hellwig 已提交
1537 1538 1539 1540

	if (mod < 0) {
		if (new_size > old_size) {
			ret = -EINVAL;
1541
			goto out_free;
C
Christoph Hellwig 已提交
1542 1543 1544
		}
		new_size = old_size - new_size;
	} else if (mod > 0) {
1545
		if (new_size > ULLONG_MAX - old_size) {
1546
			ret = -ERANGE;
1547 1548
			goto out_free;
		}
C
Christoph Hellwig 已提交
1549 1550 1551
		new_size = old_size + new_size;
	}

1552
	if (new_size < SZ_256M) {
C
Christoph Hellwig 已提交
1553
		ret = -EINVAL;
1554
		goto out_free;
C
Christoph Hellwig 已提交
1555 1556 1557
	}
	if (new_size > device->bdev->bd_inode->i_size) {
		ret = -EFBIG;
1558
		goto out_free;
C
Christoph Hellwig 已提交
1559 1560
	}

1561
	new_size = round_down(new_size, fs_info->sectorsize);
C
Christoph Hellwig 已提交
1562

1563 1564
	btrfs_info_in_rcu(fs_info, "new size for %s is %llu",
			  rcu_str_deref(device->name), new_size);
C
Christoph Hellwig 已提交
1565 1566

	if (new_size > old_size) {
1567
		trans = btrfs_start_transaction(root, 0);
1568 1569
		if (IS_ERR(trans)) {
			ret = PTR_ERR(trans);
1570
			goto out_free;
1571
		}
C
Christoph Hellwig 已提交
1572
		ret = btrfs_grow_device(trans, device, new_size);
1573
		btrfs_commit_transaction(trans);
1574
	} else if (new_size < old_size) {
C
Christoph Hellwig 已提交
1575
		ret = btrfs_shrink_device(device, new_size);
1576
	} /* equal, nothing need to do */
C
Christoph Hellwig 已提交
1577

1578
out_free:
C
Christoph Hellwig 已提交
1579
	kfree(vol_args);
1580
out:
1581
	mutex_unlock(&fs_info->volume_mutex);
1582
	clear_bit(BTRFS_FS_EXCL_OP, &fs_info->flags);
1583
	mnt_drop_write_file(file);
C
Christoph Hellwig 已提交
1584 1585 1586
	return ret;
}

S
Sage Weil 已提交
1587
static noinline int btrfs_ioctl_snap_create_transid(struct file *file,
1588
				const char *name, unsigned long fd, int subvol,
A
Arne Jansen 已提交
1589
				u64 *transid, bool readonly,
1590
				struct btrfs_qgroup_inherit *inherit)
C
Christoph Hellwig 已提交
1591 1592
{
	int namelen;
1593
	int ret = 0;
C
Christoph Hellwig 已提交
1594

1595 1596 1597
	if (!S_ISDIR(file_inode(file)->i_mode))
		return -ENOTDIR;

1598 1599 1600 1601
	ret = mnt_want_write_file(file);
	if (ret)
		goto out;

S
Sage Weil 已提交
1602 1603
	namelen = strlen(name);
	if (strchr(name, '/')) {
C
Christoph Hellwig 已提交
1604
		ret = -EINVAL;
1605
		goto out_drop_write;
C
Christoph Hellwig 已提交
1606 1607
	}

1608 1609 1610
	if (name[0] == '.' &&
	   (namelen == 1 || (name[1] == '.' && namelen == 2))) {
		ret = -EEXIST;
1611
		goto out_drop_write;
1612 1613
	}

1614
	if (subvol) {
S
Sage Weil 已提交
1615
		ret = btrfs_mksubvol(&file->f_path, name, namelen,
A
Arne Jansen 已提交
1616
				     NULL, transid, readonly, inherit);
1617
	} else {
1618
		struct fd src = fdget(fd);
1619
		struct inode *src_inode;
1620
		if (!src.file) {
1621
			ret = -EINVAL;
1622
			goto out_drop_write;
1623 1624
		}

A
Al Viro 已提交
1625 1626
		src_inode = file_inode(src.file);
		if (src_inode->i_sb != file_inode(file)->i_sb) {
J
Josef Bacik 已提交
1627
			btrfs_info(BTRFS_I(file_inode(file))->root->fs_info,
1628
				   "Snapshot src from another FS");
1629
			ret = -EXDEV;
1630 1631 1632 1633 1634 1635
		} else if (!inode_owner_or_capable(src_inode)) {
			/*
			 * Subvolume creation is not restricted, but snapshots
			 * are limited to own subvolumes only
			 */
			ret = -EPERM;
1636 1637 1638 1639
		} else {
			ret = btrfs_mksubvol(&file->f_path, name, namelen,
					     BTRFS_I(src_inode)->root,
					     transid, readonly, inherit);
1640
		}
1641
		fdput(src);
1642
	}
1643 1644
out_drop_write:
	mnt_drop_write_file(file);
C
Christoph Hellwig 已提交
1645
out:
S
Sage Weil 已提交
1646 1647 1648 1649
	return ret;
}

static noinline int btrfs_ioctl_snap_create(struct file *file,
1650
					    void __user *arg, int subvol)
S
Sage Weil 已提交
1651
{
1652
	struct btrfs_ioctl_vol_args *vol_args;
S
Sage Weil 已提交
1653 1654
	int ret;

1655 1656 1657
	if (!S_ISDIR(file_inode(file)->i_mode))
		return -ENOTDIR;

1658 1659 1660 1661
	vol_args = memdup_user(arg, sizeof(*vol_args));
	if (IS_ERR(vol_args))
		return PTR_ERR(vol_args);
	vol_args->name[BTRFS_PATH_NAME_MAX] = '\0';
S
Sage Weil 已提交
1662

1663
	ret = btrfs_ioctl_snap_create_transid(file, vol_args->name,
L
Li Zefan 已提交
1664
					      vol_args->fd, subvol,
A
Arne Jansen 已提交
1665
					      NULL, false, NULL);
1666

1667 1668 1669
	kfree(vol_args);
	return ret;
}
1670

1671 1672 1673 1674 1675 1676 1677
static noinline int btrfs_ioctl_snap_create_v2(struct file *file,
					       void __user *arg, int subvol)
{
	struct btrfs_ioctl_vol_args_v2 *vol_args;
	int ret;
	u64 transid = 0;
	u64 *ptr = NULL;
L
Li Zefan 已提交
1678
	bool readonly = false;
A
Arne Jansen 已提交
1679
	struct btrfs_qgroup_inherit *inherit = NULL;
1680

1681 1682 1683
	if (!S_ISDIR(file_inode(file)->i_mode))
		return -ENOTDIR;

1684 1685 1686 1687
	vol_args = memdup_user(arg, sizeof(*vol_args));
	if (IS_ERR(vol_args))
		return PTR_ERR(vol_args);
	vol_args->name[BTRFS_SUBVOL_NAME_MAX] = '\0';
1688

L
Li Zefan 已提交
1689
	if (vol_args->flags &
A
Arne Jansen 已提交
1690 1691
	    ~(BTRFS_SUBVOL_CREATE_ASYNC | BTRFS_SUBVOL_RDONLY |
	      BTRFS_SUBVOL_QGROUP_INHERIT)) {
L
Li Zefan 已提交
1692
		ret = -EOPNOTSUPP;
D
Dan Carpenter 已提交
1693
		goto free_args;
S
Sage Weil 已提交
1694
	}
1695 1696 1697

	if (vol_args->flags & BTRFS_SUBVOL_CREATE_ASYNC)
		ptr = &transid;
L
Li Zefan 已提交
1698 1699
	if (vol_args->flags & BTRFS_SUBVOL_RDONLY)
		readonly = true;
A
Arne Jansen 已提交
1700
	if (vol_args->flags & BTRFS_SUBVOL_QGROUP_INHERIT) {
1701
		if (vol_args->size > PAGE_SIZE) {
A
Arne Jansen 已提交
1702
			ret = -EINVAL;
D
Dan Carpenter 已提交
1703
			goto free_args;
A
Arne Jansen 已提交
1704 1705 1706 1707
		}
		inherit = memdup_user(vol_args->qgroup_inherit, vol_args->size);
		if (IS_ERR(inherit)) {
			ret = PTR_ERR(inherit);
D
Dan Carpenter 已提交
1708
			goto free_args;
A
Arne Jansen 已提交
1709 1710
		}
	}
1711 1712

	ret = btrfs_ioctl_snap_create_transid(file, vol_args->name,
A
Arne Jansen 已提交
1713
					      vol_args->fd, subvol, ptr,
1714
					      readonly, inherit);
D
Dan Carpenter 已提交
1715 1716
	if (ret)
		goto free_inherit;
1717

D
Dan Carpenter 已提交
1718 1719 1720 1721
	if (ptr && copy_to_user(arg +
				offsetof(struct btrfs_ioctl_vol_args_v2,
					transid),
				ptr, sizeof(*ptr)))
1722
		ret = -EFAULT;
D
Dan Carpenter 已提交
1723 1724

free_inherit:
A
Arne Jansen 已提交
1725
	kfree(inherit);
D
Dan Carpenter 已提交
1726 1727
free_args:
	kfree(vol_args);
C
Christoph Hellwig 已提交
1728 1729 1730
	return ret;
}

1731 1732 1733
static noinline int btrfs_ioctl_subvol_getflags(struct file *file,
						void __user *arg)
{
A
Al Viro 已提交
1734
	struct inode *inode = file_inode(file);
1735
	struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb);
1736 1737 1738 1739
	struct btrfs_root *root = BTRFS_I(inode)->root;
	int ret = 0;
	u64 flags = 0;

1740
	if (btrfs_ino(BTRFS_I(inode)) != BTRFS_FIRST_FREE_OBJECTID)
1741 1742
		return -EINVAL;

1743
	down_read(&fs_info->subvol_sem);
1744 1745
	if (btrfs_root_readonly(root))
		flags |= BTRFS_SUBVOL_RDONLY;
1746
	up_read(&fs_info->subvol_sem);
1747 1748 1749 1750 1751 1752 1753 1754 1755 1756

	if (copy_to_user(arg, &flags, sizeof(flags)))
		ret = -EFAULT;

	return ret;
}

static noinline int btrfs_ioctl_subvol_setflags(struct file *file,
					      void __user *arg)
{
A
Al Viro 已提交
1757
	struct inode *inode = file_inode(file);
1758
	struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb);
1759 1760 1761 1762 1763 1764
	struct btrfs_root *root = BTRFS_I(inode)->root;
	struct btrfs_trans_handle *trans;
	u64 root_flags;
	u64 flags;
	int ret = 0;

1765 1766 1767
	if (!inode_owner_or_capable(inode))
		return -EPERM;

1768 1769 1770
	ret = mnt_want_write_file(file);
	if (ret)
		goto out;
1771

1772
	if (btrfs_ino(BTRFS_I(inode)) != BTRFS_FIRST_FREE_OBJECTID) {
1773 1774 1775
		ret = -EINVAL;
		goto out_drop_write;
	}
1776

1777 1778 1779 1780
	if (copy_from_user(&flags, arg, sizeof(flags))) {
		ret = -EFAULT;
		goto out_drop_write;
	}
1781

1782 1783 1784 1785
	if (flags & BTRFS_SUBVOL_CREATE_ASYNC) {
		ret = -EINVAL;
		goto out_drop_write;
	}
1786

1787 1788 1789 1790
	if (flags & ~BTRFS_SUBVOL_RDONLY) {
		ret = -EOPNOTSUPP;
		goto out_drop_write;
	}
1791

1792
	down_write(&fs_info->subvol_sem);
1793 1794 1795

	/* nothing to do */
	if (!!(flags & BTRFS_SUBVOL_RDONLY) == btrfs_root_readonly(root))
1796
		goto out_drop_sem;
1797 1798

	root_flags = btrfs_root_flags(&root->root_item);
1799
	if (flags & BTRFS_SUBVOL_RDONLY) {
1800 1801
		btrfs_set_root_flags(&root->root_item,
				     root_flags | BTRFS_ROOT_SUBVOL_RDONLY);
1802 1803 1804 1805 1806 1807 1808 1809
	} else {
		/*
		 * Block RO -> RW transition if this subvolume is involved in
		 * send
		 */
		spin_lock(&root->root_item_lock);
		if (root->send_in_progress == 0) {
			btrfs_set_root_flags(&root->root_item,
1810
				     root_flags & ~BTRFS_ROOT_SUBVOL_RDONLY);
1811 1812 1813
			spin_unlock(&root->root_item_lock);
		} else {
			spin_unlock(&root->root_item_lock);
1814 1815 1816
			btrfs_warn(fs_info,
				   "Attempt to set subvolume %llu read-write during send",
				   root->root_key.objectid);
1817 1818 1819 1820
			ret = -EPERM;
			goto out_drop_sem;
		}
	}
1821 1822 1823 1824 1825 1826 1827

	trans = btrfs_start_transaction(root, 1);
	if (IS_ERR(trans)) {
		ret = PTR_ERR(trans);
		goto out_reset;
	}

1828
	ret = btrfs_update_root(trans, fs_info->tree_root,
1829 1830
				&root->root_key, &root->root_item);

1831
	btrfs_commit_transaction(trans);
1832 1833 1834
out_reset:
	if (ret)
		btrfs_set_root_flags(&root->root_item, root_flags);
1835
out_drop_sem:
1836
	up_write(&fs_info->subvol_sem);
1837 1838 1839
out_drop_write:
	mnt_drop_write_file(file);
out:
1840 1841 1842
	return ret;
}

1843 1844 1845 1846 1847
/*
 * helper to check if the subvolume references other subvolumes
 */
static noinline int may_destroy_subvol(struct btrfs_root *root)
{
1848
	struct btrfs_fs_info *fs_info = root->fs_info;
1849
	struct btrfs_path *path;
1850
	struct btrfs_dir_item *di;
1851
	struct btrfs_key key;
1852
	u64 dir_id;
1853 1854 1855 1856 1857 1858
	int ret;

	path = btrfs_alloc_path();
	if (!path)
		return -ENOMEM;

1859
	/* Make sure this root isn't set as the default subvol */
1860 1861
	dir_id = btrfs_super_root_dir(fs_info->super_copy);
	di = btrfs_lookup_dir_item(NULL, fs_info->tree_root, path,
1862 1863 1864 1865
				   dir_id, "default", 7, 0);
	if (di && !IS_ERR(di)) {
		btrfs_dir_item_key_to_cpu(path->nodes[0], di, &key);
		if (key.objectid == root->root_key.objectid) {
1866
			ret = -EPERM;
1867
			btrfs_err(fs_info,
J
Jeff Mahoney 已提交
1868 1869
				  "deleting default subvolume %llu is not allowed",
				  key.objectid);
1870 1871 1872 1873 1874
			goto out;
		}
		btrfs_release_path(path);
	}

1875 1876 1877 1878
	key.objectid = root->root_key.objectid;
	key.type = BTRFS_ROOT_REF_KEY;
	key.offset = (u64)-1;

1879
	ret = btrfs_search_slot(NULL, fs_info->tree_root, &key, path, 0, 0);
1880 1881 1882 1883 1884 1885 1886 1887 1888 1889 1890 1891 1892 1893 1894 1895 1896
	if (ret < 0)
		goto out;
	BUG_ON(ret == 0);

	ret = 0;
	if (path->slots[0] > 0) {
		path->slots[0]--;
		btrfs_item_key_to_cpu(path->nodes[0], &key, path->slots[0]);
		if (key.objectid == root->root_key.objectid &&
		    key.type == BTRFS_ROOT_REF_KEY)
			ret = -ENOTEMPTY;
	}
out:
	btrfs_free_path(path);
	return ret;
}

1897 1898 1899
static noinline int key_in_sk(struct btrfs_key *key,
			      struct btrfs_ioctl_search_key *sk)
{
1900 1901 1902 1903 1904 1905 1906 1907 1908
	struct btrfs_key test;
	int ret;

	test.objectid = sk->min_objectid;
	test.type = sk->min_type;
	test.offset = sk->min_offset;

	ret = btrfs_comp_cpu_keys(key, &test);
	if (ret < 0)
1909
		return 0;
1910 1911 1912 1913 1914 1915 1916

	test.objectid = sk->max_objectid;
	test.type = sk->max_type;
	test.offset = sk->max_offset;

	ret = btrfs_comp_cpu_keys(key, &test);
	if (ret > 0)
1917 1918 1919 1920
		return 0;
	return 1;
}

1921
static noinline int copy_to_sk(struct btrfs_path *path,
1922 1923
			       struct btrfs_key *key,
			       struct btrfs_ioctl_search_key *sk,
1924
			       size_t *buf_size,
1925
			       char __user *ubuf,
1926 1927 1928 1929 1930 1931
			       unsigned long *sk_offset,
			       int *num_found)
{
	u64 found_transid;
	struct extent_buffer *leaf;
	struct btrfs_ioctl_search_header sh;
1932
	struct btrfs_key test;
1933 1934 1935 1936 1937 1938 1939 1940 1941 1942 1943 1944 1945 1946 1947 1948 1949 1950 1951 1952 1953
	unsigned long item_off;
	unsigned long item_len;
	int nritems;
	int i;
	int slot;
	int ret = 0;

	leaf = path->nodes[0];
	slot = path->slots[0];
	nritems = btrfs_header_nritems(leaf);

	if (btrfs_header_generation(leaf) > sk->max_transid) {
		i = nritems;
		goto advance_key;
	}
	found_transid = btrfs_header_generation(leaf);

	for (i = slot; i < nritems; i++) {
		item_off = btrfs_item_ptr_offset(leaf, i);
		item_len = btrfs_item_size_nr(leaf, i);

1954 1955 1956 1957
		btrfs_item_key_to_cpu(leaf, key, i);
		if (!key_in_sk(key, sk))
			continue;

1958
		if (sizeof(sh) + item_len > *buf_size) {
1959 1960 1961 1962 1963 1964 1965 1966 1967 1968
			if (*num_found) {
				ret = 1;
				goto out;
			}

			/*
			 * return one empty item back for v1, which does not
			 * handle -EOVERFLOW
			 */

1969
			*buf_size = sizeof(sh) + item_len;
1970
			item_len = 0;
1971 1972
			ret = -EOVERFLOW;
		}
1973

1974
		if (sizeof(sh) + item_len + *sk_offset > *buf_size) {
1975
			ret = 1;
1976
			goto out;
1977 1978 1979 1980 1981 1982 1983 1984 1985
		}

		sh.objectid = key->objectid;
		sh.offset = key->offset;
		sh.type = key->type;
		sh.len = item_len;
		sh.transid = found_transid;

		/* copy search result header */
1986 1987 1988 1989 1990
		if (copy_to_user(ubuf + *sk_offset, &sh, sizeof(sh))) {
			ret = -EFAULT;
			goto out;
		}

1991 1992 1993
		*sk_offset += sizeof(sh);

		if (item_len) {
1994
			char __user *up = ubuf + *sk_offset;
1995
			/* copy the item */
1996 1997 1998 1999 2000 2001
			if (read_extent_buffer_to_user(leaf, up,
						       item_off, item_len)) {
				ret = -EFAULT;
				goto out;
			}

2002 2003
			*sk_offset += item_len;
		}
2004
		(*num_found)++;
2005

2006 2007 2008
		if (ret) /* -EOVERFLOW from above */
			goto out;

2009 2010 2011 2012
		if (*num_found >= sk->nr_items) {
			ret = 1;
			goto out;
		}
2013 2014
	}
advance_key:
2015
	ret = 0;
2016 2017 2018 2019 2020 2021
	test.objectid = sk->max_objectid;
	test.type = sk->max_type;
	test.offset = sk->max_offset;
	if (btrfs_comp_cpu_keys(key, &test) >= 0)
		ret = 1;
	else if (key->offset < (u64)-1)
2022
		key->offset++;
2023
	else if (key->type < (u8)-1) {
2024
		key->offset = 0;
2025
		key->type++;
2026
	} else if (key->objectid < (u64)-1) {
2027 2028
		key->offset = 0;
		key->type = 0;
2029
		key->objectid++;
2030 2031
	} else
		ret = 1;
2032
out:
2033 2034 2035 2036 2037 2038 2039 2040 2041
	/*
	 *  0: all items from this leaf copied, continue with next
	 *  1: * more items can be copied, but unused buffer is too small
	 *     * all items were found
	 *     Either way, it will stops the loop which iterates to the next
	 *     leaf
	 *  -EOVERFLOW: item was to large for buffer
	 *  -EFAULT: could not copy extent buffer back to userspace
	 */
2042 2043 2044 2045
	return ret;
}

static noinline int search_ioctl(struct inode *inode,
2046
				 struct btrfs_ioctl_search_key *sk,
2047
				 size_t *buf_size,
2048
				 char __user *ubuf)
2049
{
2050
	struct btrfs_fs_info *info = btrfs_sb(inode->i_sb);
2051 2052 2053 2054 2055 2056 2057
	struct btrfs_root *root;
	struct btrfs_key key;
	struct btrfs_path *path;
	int ret;
	int num_found = 0;
	unsigned long sk_offset = 0;

2058 2059
	if (*buf_size < sizeof(struct btrfs_ioctl_search_header)) {
		*buf_size = sizeof(struct btrfs_ioctl_search_header);
2060
		return -EOVERFLOW;
2061
	}
2062

2063 2064 2065 2066 2067 2068 2069 2070 2071 2072 2073 2074 2075 2076 2077 2078 2079 2080 2081 2082 2083 2084
	path = btrfs_alloc_path();
	if (!path)
		return -ENOMEM;

	if (sk->tree_id == 0) {
		/* search the root of the inode that was passed */
		root = BTRFS_I(inode)->root;
	} else {
		key.objectid = sk->tree_id;
		key.type = BTRFS_ROOT_ITEM_KEY;
		key.offset = (u64)-1;
		root = btrfs_read_fs_root_no_name(info, &key);
		if (IS_ERR(root)) {
			btrfs_free_path(path);
			return -ENOENT;
		}
	}

	key.objectid = sk->min_objectid;
	key.type = sk->min_type;
	key.offset = sk->min_offset;

2085
	while (1) {
2086
		ret = btrfs_search_forward(root, &key, path, sk->min_transid);
2087 2088 2089 2090 2091
		if (ret != 0) {
			if (ret > 0)
				ret = 0;
			goto err;
		}
2092
		ret = copy_to_sk(path, &key, sk, buf_size, ubuf,
2093
				 &sk_offset, &num_found);
2094
		btrfs_release_path(path);
2095
		if (ret)
2096 2097 2098
			break;

	}
2099 2100
	if (ret > 0)
		ret = 0;
2101 2102 2103 2104 2105 2106 2107 2108 2109
err:
	sk->nr_items = num_found;
	btrfs_free_path(path);
	return ret;
}

static noinline int btrfs_ioctl_tree_search(struct file *file,
					   void __user *argp)
{
2110 2111
	struct btrfs_ioctl_search_args __user *uargs;
	struct btrfs_ioctl_search_key sk;
2112 2113 2114
	struct inode *inode;
	int ret;
	size_t buf_size;
2115 2116 2117 2118

	if (!capable(CAP_SYS_ADMIN))
		return -EPERM;

2119 2120 2121 2122
	uargs = (struct btrfs_ioctl_search_args __user *)argp;

	if (copy_from_user(&sk, &uargs->key, sizeof(sk)))
		return -EFAULT;
2123

2124
	buf_size = sizeof(uargs->buf);
2125

A
Al Viro 已提交
2126
	inode = file_inode(file);
2127
	ret = search_ioctl(inode, &sk, &buf_size, uargs->buf);
2128 2129 2130 2131 2132 2133 2134 2135

	/*
	 * In the origin implementation an overflow is handled by returning a
	 * search header with a len of zero, so reset ret.
	 */
	if (ret == -EOVERFLOW)
		ret = 0;

2136
	if (ret == 0 && copy_to_user(&uargs->key, &sk, sizeof(sk)))
2137 2138 2139 2140
		ret = -EFAULT;
	return ret;
}

G
Gerhard Heift 已提交
2141 2142 2143 2144 2145 2146 2147 2148
static noinline int btrfs_ioctl_tree_search_v2(struct file *file,
					       void __user *argp)
{
	struct btrfs_ioctl_search_args_v2 __user *uarg;
	struct btrfs_ioctl_search_args_v2 args;
	struct inode *inode;
	int ret;
	size_t buf_size;
2149
	const size_t buf_limit = SZ_16M;
G
Gerhard Heift 已提交
2150 2151 2152 2153 2154 2155 2156 2157 2158 2159 2160 2161 2162 2163 2164 2165 2166 2167 2168 2169 2170 2171 2172 2173

	if (!capable(CAP_SYS_ADMIN))
		return -EPERM;

	/* copy search header and buffer size */
	uarg = (struct btrfs_ioctl_search_args_v2 __user *)argp;
	if (copy_from_user(&args, uarg, sizeof(args)))
		return -EFAULT;

	buf_size = args.buf_size;

	/* limit result size to 16MB */
	if (buf_size > buf_limit)
		buf_size = buf_limit;

	inode = file_inode(file);
	ret = search_ioctl(inode, &args.key, &buf_size,
			   (char *)(&uarg->buf[0]));
	if (ret == 0 && copy_to_user(&uarg->key, &args.key, sizeof(args.key)))
		ret = -EFAULT;
	else if (ret == -EOVERFLOW &&
		copy_to_user(&uarg->buf_size, &buf_size, sizeof(buf_size)))
		ret = -EFAULT;

2174 2175 2176
	return ret;
}

2177
/*
2178 2179 2180
 * Search INODE_REFs to identify path name of 'dirid' directory
 * in a 'tree_id' tree. and sets path name to 'name'.
 */
2181 2182 2183 2184 2185
static noinline int btrfs_search_path_in_tree(struct btrfs_fs_info *info,
				u64 tree_id, u64 dirid, char *name)
{
	struct btrfs_root *root;
	struct btrfs_key key;
2186
	char *ptr;
2187 2188 2189 2190 2191 2192 2193 2194 2195 2196 2197 2198 2199 2200 2201 2202 2203
	int ret = -1;
	int slot;
	int len;
	int total_len = 0;
	struct btrfs_inode_ref *iref;
	struct extent_buffer *l;
	struct btrfs_path *path;

	if (dirid == BTRFS_FIRST_FREE_OBJECTID) {
		name[0]='\0';
		return 0;
	}

	path = btrfs_alloc_path();
	if (!path)
		return -ENOMEM;

2204
	ptr = &name[BTRFS_INO_LOOKUP_PATH_MAX];
2205 2206 2207 2208 2209 2210

	key.objectid = tree_id;
	key.type = BTRFS_ROOT_ITEM_KEY;
	key.offset = (u64)-1;
	root = btrfs_read_fs_root_no_name(info, &key);
	if (IS_ERR(root)) {
2211
		btrfs_err(info, "could not find root %llu", tree_id);
2212 2213
		ret = -ENOENT;
		goto out;
2214 2215 2216 2217
	}

	key.objectid = dirid;
	key.type = BTRFS_INODE_REF_KEY;
2218
	key.offset = (u64)-1;
2219

2220
	while (1) {
2221 2222 2223
		ret = btrfs_search_slot(NULL, root, &key, path, 0, 0);
		if (ret < 0)
			goto out;
2224 2225 2226 2227 2228 2229 2230 2231 2232 2233
		else if (ret > 0) {
			ret = btrfs_previous_item(root, path, dirid,
						  BTRFS_INODE_REF_KEY);
			if (ret < 0)
				goto out;
			else if (ret > 0) {
				ret = -ENOENT;
				goto out;
			}
		}
2234 2235 2236 2237 2238 2239 2240 2241 2242

		l = path->nodes[0];
		slot = path->slots[0];
		btrfs_item_key_to_cpu(l, &key, slot);

		iref = btrfs_item_ptr(l, slot, struct btrfs_inode_ref);
		len = btrfs_inode_ref_name_len(l, iref);
		ptr -= len + 1;
		total_len += len + 1;
2243 2244
		if (ptr < name) {
			ret = -ENAMETOOLONG;
2245
			goto out;
2246
		}
2247 2248

		*(ptr + len) = '/';
2249
		read_extent_buffer(l, ptr, (unsigned long)(iref + 1), len);
2250 2251 2252 2253

		if (key.offset == BTRFS_FIRST_FREE_OBJECTID)
			break;

2254
		btrfs_release_path(path);
2255
		key.objectid = key.offset;
2256
		key.offset = (u64)-1;
2257 2258
		dirid = key.objectid;
	}
2259
	memmove(name, ptr, total_len);
2260
	name[total_len] = '\0';
2261 2262 2263
	ret = 0;
out:
	btrfs_free_path(path);
2264 2265 2266 2267 2268 2269 2270 2271
	return ret;
}

static noinline int btrfs_ioctl_ino_lookup(struct file *file,
					   void __user *argp)
{
	 struct btrfs_ioctl_ino_lookup_args *args;
	 struct inode *inode;
2272
	int ret = 0;
2273

J
Julia Lawall 已提交
2274 2275 2276
	args = memdup_user(argp, sizeof(*args));
	if (IS_ERR(args))
		return PTR_ERR(args);
2277

A
Al Viro 已提交
2278
	inode = file_inode(file);
2279

2280 2281 2282 2283
	/*
	 * Unprivileged query to obtain the containing subvolume root id. The
	 * path is reset so it's consistent with btrfs_search_path_in_tree.
	 */
2284 2285 2286
	if (args->treeid == 0)
		args->treeid = BTRFS_I(inode)->root->root_key.objectid;

2287 2288 2289 2290 2291 2292 2293 2294 2295 2296
	if (args->objectid == BTRFS_FIRST_FREE_OBJECTID) {
		args->name[0] = 0;
		goto out;
	}

	if (!capable(CAP_SYS_ADMIN)) {
		ret = -EPERM;
		goto out;
	}

2297 2298 2299 2300
	ret = btrfs_search_path_in_tree(BTRFS_I(inode)->root->fs_info,
					args->treeid, args->objectid,
					args->name);

2301
out:
2302 2303 2304 2305
	if (ret == 0 && copy_to_user(argp, args, sizeof(*args)))
		ret = -EFAULT;

	kfree(args);
2306 2307 2308
	return ret;
}

2309 2310 2311
static noinline int btrfs_ioctl_snap_destroy(struct file *file,
					     void __user *arg)
{
A
Al Viro 已提交
2312
	struct dentry *parent = file->f_path.dentry;
2313
	struct btrfs_fs_info *fs_info = btrfs_sb(parent->d_sb);
2314
	struct dentry *dentry;
2315
	struct inode *dir = d_inode(parent);
2316 2317 2318 2319 2320
	struct inode *inode;
	struct btrfs_root *root = BTRFS_I(dir)->root;
	struct btrfs_root *dest = NULL;
	struct btrfs_ioctl_vol_args *vol_args;
	struct btrfs_trans_handle *trans;
2321
	struct btrfs_block_rsv block_rsv;
2322
	u64 root_flags;
2323
	u64 qgroup_reserved;
2324 2325 2326 2327
	int namelen;
	int ret;
	int err = 0;

2328 2329 2330
	if (!S_ISDIR(dir->i_mode))
		return -ENOTDIR;

2331 2332 2333 2334 2335 2336 2337 2338 2339 2340 2341 2342
	vol_args = memdup_user(arg, sizeof(*vol_args));
	if (IS_ERR(vol_args))
		return PTR_ERR(vol_args);

	vol_args->name[BTRFS_PATH_NAME_MAX] = '\0';
	namelen = strlen(vol_args->name);
	if (strchr(vol_args->name, '/') ||
	    strncmp(vol_args->name, "..", namelen) == 0) {
		err = -EINVAL;
		goto out;
	}

2343
	err = mnt_want_write_file(file);
2344 2345 2346
	if (err)
		goto out;

2347

2348 2349 2350
	err = down_write_killable_nested(&dir->i_rwsem, I_MUTEX_PARENT);
	if (err == -EINTR)
		goto out_drop_write;
2351 2352 2353 2354 2355 2356
	dentry = lookup_one_len(vol_args->name, parent, namelen);
	if (IS_ERR(dentry)) {
		err = PTR_ERR(dentry);
		goto out_unlock_dir;
	}

2357
	if (d_really_is_negative(dentry)) {
2358 2359 2360 2361
		err = -ENOENT;
		goto out_dput;
	}

2362
	inode = d_inode(dentry);
2363
	dest = BTRFS_I(inode)->root;
2364
	if (!capable(CAP_SYS_ADMIN)) {
2365 2366 2367 2368 2369 2370 2371 2372 2373 2374 2375 2376 2377 2378
		/*
		 * Regular user.  Only allow this with a special mount
		 * option, when the user has write+exec access to the
		 * subvol root, and when rmdir(2) would have been
		 * allowed.
		 *
		 * Note that this is _not_ check that the subvol is
		 * empty or doesn't contain data that we wouldn't
		 * otherwise be able to delete.
		 *
		 * Users who want to delete empty subvols should try
		 * rmdir(2).
		 */
		err = -EPERM;
2379
		if (!btrfs_test_opt(fs_info, USER_SUBVOL_RM_ALLOWED))
2380 2381 2382 2383 2384 2385 2386 2387 2388 2389 2390 2391 2392 2393 2394 2395 2396 2397
			goto out_dput;

		/*
		 * Do not allow deletion if the parent dir is the same
		 * as the dir to be deleted.  That means the ioctl
		 * must be called on the dentry referencing the root
		 * of the subvol, not a random directory contained
		 * within it.
		 */
		err = -EINVAL;
		if (root == dest)
			goto out_dput;

		err = inode_permission(inode, MAY_WRITE | MAY_EXEC);
		if (err)
			goto out_dput;
	}

2398 2399 2400 2401 2402
	/* check if subvolume may be deleted by a user */
	err = btrfs_may_delete(dir, dentry, 1);
	if (err)
		goto out_dput;

2403
	if (btrfs_ino(BTRFS_I(inode)) != BTRFS_FIRST_FREE_OBJECTID) {
2404 2405 2406 2407
		err = -EINVAL;
		goto out_dput;
	}

A
Al Viro 已提交
2408
	inode_lock(inode);
2409 2410 2411 2412 2413 2414 2415

	/*
	 * Don't allow to delete a subvolume with send in progress. This is
	 * inside the i_mutex so the error handling that has to drop the bit
	 * again is not run concurrently.
	 */
	spin_lock(&dest->root_item_lock);
2416 2417 2418
	root_flags = btrfs_root_flags(&dest->root_item);
	if (dest->send_in_progress == 0) {
		btrfs_set_root_flags(&dest->root_item,
2419 2420 2421 2422
				root_flags | BTRFS_ROOT_SUBVOL_DEAD);
		spin_unlock(&dest->root_item_lock);
	} else {
		spin_unlock(&dest->root_item_lock);
2423 2424 2425
		btrfs_warn(fs_info,
			   "Attempt to delete subvolume %llu during send",
			   dest->root_key.objectid);
2426
		err = -EPERM;
2427
		goto out_unlock_inode;
2428 2429
	}

2430
	down_write(&fs_info->subvol_sem);
2431 2432 2433 2434 2435

	err = may_destroy_subvol(dest);
	if (err)
		goto out_up_write;

2436 2437 2438 2439 2440 2441
	btrfs_init_block_rsv(&block_rsv, BTRFS_BLOCK_RSV_TEMP);
	/*
	 * One for dir inode, two for dir entries, two for root
	 * ref/backref.
	 */
	err = btrfs_subvolume_reserve_metadata(root, &block_rsv,
2442
					       5, &qgroup_reserved, true);
2443 2444 2445
	if (err)
		goto out_up_write;

2446 2447 2448
	trans = btrfs_start_transaction(root, 0);
	if (IS_ERR(trans)) {
		err = PTR_ERR(trans);
2449
		goto out_release;
2450
	}
2451 2452
	trans->block_rsv = &block_rsv;
	trans->bytes_reserved = block_rsv.size;
2453

2454
	btrfs_record_snapshot_destroy(trans, BTRFS_I(dir));
2455

2456 2457 2458 2459
	ret = btrfs_unlink_subvol(trans, root, dir,
				dest->root_key.objectid,
				dentry->d_name.name,
				dentry->d_name.len);
2460 2461
	if (ret) {
		err = ret;
2462
		btrfs_abort_transaction(trans, ret);
2463 2464
		goto out_end_trans;
	}
2465 2466 2467 2468 2469 2470 2471 2472

	btrfs_record_root_in_trans(trans, dest);

	memset(&dest->root_item.drop_progress, 0,
		sizeof(dest->root_item.drop_progress));
	dest->root_item.drop_level = 0;
	btrfs_set_root_refs(&dest->root_item, 0);

2473
	if (!test_and_set_bit(BTRFS_ROOT_ORPHAN_ITEM_INSERTED, &dest->state)) {
2474
		ret = btrfs_insert_orphan_item(trans,
2475
					fs_info->tree_root,
2476
					dest->root_key.objectid);
2477
		if (ret) {
2478
			btrfs_abort_transaction(trans, ret);
2479 2480 2481
			err = ret;
			goto out_end_trans;
		}
2482
	}
2483

2484
	ret = btrfs_uuid_tree_rem(trans, fs_info, dest->root_item.uuid,
2485
				  BTRFS_UUID_KEY_SUBVOL,
2486 2487
				  dest->root_key.objectid);
	if (ret && ret != -ENOENT) {
2488
		btrfs_abort_transaction(trans, ret);
2489 2490 2491 2492
		err = ret;
		goto out_end_trans;
	}
	if (!btrfs_is_empty_uuid(dest->root_item.received_uuid)) {
2493
		ret = btrfs_uuid_tree_rem(trans, fs_info,
2494 2495 2496 2497
					  dest->root_item.received_uuid,
					  BTRFS_UUID_KEY_RECEIVED_SUBVOL,
					  dest->root_key.objectid);
		if (ret && ret != -ENOENT) {
2498
			btrfs_abort_transaction(trans, ret);
2499 2500 2501 2502 2503
			err = ret;
			goto out_end_trans;
		}
	}

2504
out_end_trans:
2505 2506
	trans->block_rsv = NULL;
	trans->bytes_reserved = 0;
2507
	ret = btrfs_end_transaction(trans);
2508 2509
	if (ret && !err)
		err = ret;
2510
	inode->i_flags |= S_DEAD;
2511
out_release:
2512
	btrfs_subvolume_release_metadata(fs_info, &block_rsv);
2513
out_up_write:
2514
	up_write(&fs_info->subvol_sem);
2515 2516
	if (err) {
		spin_lock(&dest->root_item_lock);
2517 2518
		root_flags = btrfs_root_flags(&dest->root_item);
		btrfs_set_root_flags(&dest->root_item,
2519 2520 2521
				root_flags & ~BTRFS_ROOT_SUBVOL_DEAD);
		spin_unlock(&dest->root_item_lock);
	}
2522
out_unlock_inode:
A
Al Viro 已提交
2523
	inode_unlock(inode);
2524
	if (!err) {
2525
		d_invalidate(dentry);
2526 2527
		btrfs_invalidate_inodes(dest);
		d_delete(dentry);
2528
		ASSERT(dest->send_in_progress == 0);
2529 2530

		/* the last ref */
2531 2532 2533
		if (dest->ino_cache_inode) {
			iput(dest->ino_cache_inode);
			dest->ino_cache_inode = NULL;
2534
		}
2535 2536 2537 2538
	}
out_dput:
	dput(dentry);
out_unlock_dir:
A
Al Viro 已提交
2539
	inode_unlock(dir);
2540
out_drop_write:
A
Al Viro 已提交
2541
	mnt_drop_write_file(file);
2542 2543 2544 2545 2546
out:
	kfree(vol_args);
	return err;
}

C
Chris Mason 已提交
2547
static int btrfs_ioctl_defrag(struct file *file, void __user *argp)
C
Christoph Hellwig 已提交
2548
{
A
Al Viro 已提交
2549
	struct inode *inode = file_inode(file);
C
Christoph Hellwig 已提交
2550
	struct btrfs_root *root = BTRFS_I(inode)->root;
C
Chris Mason 已提交
2551
	struct btrfs_ioctl_defrag_range_args *range;
Y
Yan Zheng 已提交
2552 2553
	int ret;

2554 2555 2556
	ret = mnt_want_write_file(file);
	if (ret)
		return ret;
L
Li Zefan 已提交
2557

2558 2559 2560
	if (btrfs_root_readonly(root)) {
		ret = -EROFS;
		goto out;
2561
	}
C
Christoph Hellwig 已提交
2562 2563 2564

	switch (inode->i_mode & S_IFMT) {
	case S_IFDIR:
2565 2566 2567 2568
		if (!capable(CAP_SYS_ADMIN)) {
			ret = -EPERM;
			goto out;
		}
2569
		ret = btrfs_defrag_root(root);
C
Christoph Hellwig 已提交
2570 2571
		break;
	case S_IFREG:
2572 2573 2574 2575
		if (!(file->f_mode & FMODE_WRITE)) {
			ret = -EINVAL;
			goto out;
		}
C
Chris Mason 已提交
2576 2577 2578 2579 2580 2581 2582 2583 2584 2585 2586 2587

		range = kzalloc(sizeof(*range), GFP_KERNEL);
		if (!range) {
			ret = -ENOMEM;
			goto out;
		}

		if (argp) {
			if (copy_from_user(range, argp,
					   sizeof(*range))) {
				ret = -EFAULT;
				kfree(range);
2588
				goto out;
C
Chris Mason 已提交
2589 2590 2591 2592 2593 2594 2595 2596 2597 2598
			}
			/* compression requires us to start the IO */
			if ((range->flags & BTRFS_DEFRAG_RANGE_COMPRESS)) {
				range->flags |= BTRFS_DEFRAG_RANGE_START_IO;
				range->extent_thresh = (u32)-1;
			}
		} else {
			/* the rest are all set to zero by kzalloc */
			range->len = (u64)-1;
		}
A
Al Viro 已提交
2599
		ret = btrfs_defrag_file(file_inode(file), file,
C
Chris Mason 已提交
2600 2601 2602
					range, 0, 0);
		if (ret > 0)
			ret = 0;
C
Chris Mason 已提交
2603
		kfree(range);
C
Christoph Hellwig 已提交
2604
		break;
2605 2606
	default:
		ret = -EINVAL;
C
Christoph Hellwig 已提交
2607
	}
2608
out:
2609
	mnt_drop_write_file(file);
2610
	return ret;
C
Christoph Hellwig 已提交
2611 2612
}

2613
static long btrfs_ioctl_add_dev(struct btrfs_fs_info *fs_info, void __user *arg)
C
Christoph Hellwig 已提交
2614 2615 2616 2617
{
	struct btrfs_ioctl_vol_args *vol_args;
	int ret;

2618 2619 2620
	if (!capable(CAP_SYS_ADMIN))
		return -EPERM;

2621
	if (test_and_set_bit(BTRFS_FS_EXCL_OP, &fs_info->flags))
2622
		return BTRFS_ERROR_DEV_EXCL_RUN_IN_PROGRESS;
2623

2624
	mutex_lock(&fs_info->volume_mutex);
L
Li Zefan 已提交
2625
	vol_args = memdup_user(arg, sizeof(*vol_args));
2626 2627 2628 2629
	if (IS_ERR(vol_args)) {
		ret = PTR_ERR(vol_args);
		goto out;
	}
C
Christoph Hellwig 已提交
2630

2631
	vol_args->name[BTRFS_PATH_NAME_MAX] = '\0';
2632
	ret = btrfs_init_new_device(fs_info, vol_args->name);
C
Christoph Hellwig 已提交
2633

A
Anand Jain 已提交
2634
	if (!ret)
2635
		btrfs_info(fs_info, "disk added %s", vol_args->name);
A
Anand Jain 已提交
2636

C
Christoph Hellwig 已提交
2637
	kfree(vol_args);
2638
out:
2639
	mutex_unlock(&fs_info->volume_mutex);
2640
	clear_bit(BTRFS_FS_EXCL_OP, &fs_info->flags);
C
Christoph Hellwig 已提交
2641 2642 2643
	return ret;
}

2644
static long btrfs_ioctl_rm_dev_v2(struct file *file, void __user *arg)
C
Christoph Hellwig 已提交
2645
{
2646 2647
	struct inode *inode = file_inode(file);
	struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb);
2648
	struct btrfs_ioctl_vol_args_v2 *vol_args;
C
Christoph Hellwig 已提交
2649 2650
	int ret;

2651 2652 2653
	if (!capable(CAP_SYS_ADMIN))
		return -EPERM;

2654 2655 2656
	ret = mnt_want_write_file(file);
	if (ret)
		return ret;
Y
Yan Zheng 已提交
2657

L
Li Zefan 已提交
2658
	vol_args = memdup_user(arg, sizeof(*vol_args));
2659 2660
	if (IS_ERR(vol_args)) {
		ret = PTR_ERR(vol_args);
D
Dan Carpenter 已提交
2661
		goto err_drop;
2662
	}
C
Christoph Hellwig 已提交
2663

2664
	/* Check for compatibility reject unknown flags */
2665 2666
	if (vol_args->flags & ~BTRFS_VOL_ARG_V2_FLAGS_SUPPORTED)
		return -EOPNOTSUPP;
C
Christoph Hellwig 已提交
2667

2668
	if (test_and_set_bit(BTRFS_FS_EXCL_OP, &fs_info->flags)) {
2669 2670 2671 2672
		ret = BTRFS_ERROR_DEV_EXCL_RUN_IN_PROGRESS;
		goto out;
	}

2673
	mutex_lock(&fs_info->volume_mutex);
2674
	if (vol_args->flags & BTRFS_DEVICE_SPEC_BY_ID) {
2675
		ret = btrfs_rm_device(fs_info, NULL, vol_args->devid);
2676 2677
	} else {
		vol_args->name[BTRFS_SUBVOL_NAME_MAX] = '\0';
2678
		ret = btrfs_rm_device(fs_info, vol_args->name, 0);
2679
	}
2680
	mutex_unlock(&fs_info->volume_mutex);
2681
	clear_bit(BTRFS_FS_EXCL_OP, &fs_info->flags);
2682

2683
	if (!ret) {
2684
		if (vol_args->flags & BTRFS_DEVICE_SPEC_BY_ID)
2685
			btrfs_info(fs_info, "device deleted: id %llu",
2686 2687
					vol_args->devid);
		else
2688
			btrfs_info(fs_info, "device deleted: %s",
2689 2690
					vol_args->name);
	}
2691 2692
out:
	kfree(vol_args);
D
Dan Carpenter 已提交
2693
err_drop:
2694
	mnt_drop_write_file(file);
C
Christoph Hellwig 已提交
2695 2696 2697
	return ret;
}

2698
static long btrfs_ioctl_rm_dev(struct file *file, void __user *arg)
C
Christoph Hellwig 已提交
2699
{
2700 2701
	struct inode *inode = file_inode(file);
	struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb);
C
Christoph Hellwig 已提交
2702 2703 2704
	struct btrfs_ioctl_vol_args *vol_args;
	int ret;

2705 2706 2707
	if (!capable(CAP_SYS_ADMIN))
		return -EPERM;

2708 2709 2710
	ret = mnt_want_write_file(file);
	if (ret)
		return ret;
Y
Yan Zheng 已提交
2711

2712
	if (test_and_set_bit(BTRFS_FS_EXCL_OP, &fs_info->flags)) {
2713
		ret = BTRFS_ERROR_DEV_EXCL_RUN_IN_PROGRESS;
2714 2715 2716 2717 2718 2719
		goto out_drop_write;
	}

	vol_args = memdup_user(arg, sizeof(*vol_args));
	if (IS_ERR(vol_args)) {
		ret = PTR_ERR(vol_args);
2720 2721 2722
		goto out;
	}

2723
	vol_args->name[BTRFS_PATH_NAME_MAX] = '\0';
2724
	mutex_lock(&fs_info->volume_mutex);
2725
	ret = btrfs_rm_device(fs_info, vol_args->name, 0);
2726
	mutex_unlock(&fs_info->volume_mutex);
2727

2728
	if (!ret)
2729
		btrfs_info(fs_info, "disk deleted %s", vol_args->name);
2730
	kfree(vol_args);
2731
out:
2732
	clear_bit(BTRFS_FS_EXCL_OP, &fs_info->flags);
2733
out_drop_write:
2734
	mnt_drop_write_file(file);
2735

C
Christoph Hellwig 已提交
2736 2737 2738
	return ret;
}

2739 2740
static long btrfs_ioctl_fs_info(struct btrfs_fs_info *fs_info,
				void __user *arg)
J
Jan Schmidt 已提交
2741
{
2742
	struct btrfs_ioctl_fs_info_args *fi_args;
J
Jan Schmidt 已提交
2743
	struct btrfs_device *device;
2744
	struct btrfs_fs_devices *fs_devices = fs_info->fs_devices;
2745
	int ret = 0;
J
Jan Schmidt 已提交
2746

2747 2748 2749 2750
	fi_args = kzalloc(sizeof(*fi_args), GFP_KERNEL);
	if (!fi_args)
		return -ENOMEM;

2751
	mutex_lock(&fs_devices->device_list_mutex);
2752
	fi_args->num_devices = fs_devices->num_devices;
2753
	memcpy(&fi_args->fsid, fs_info->fsid, sizeof(fi_args->fsid));
J
Jan Schmidt 已提交
2754

2755
	list_for_each_entry(device, &fs_devices->devices, dev_list) {
2756 2757
		if (device->devid > fi_args->max_id)
			fi_args->max_id = device->devid;
J
Jan Schmidt 已提交
2758 2759 2760
	}
	mutex_unlock(&fs_devices->device_list_mutex);

2761 2762 2763
	fi_args->nodesize = fs_info->nodesize;
	fi_args->sectorsize = fs_info->sectorsize;
	fi_args->clone_alignment = fs_info->sectorsize;
2764

2765 2766
	if (copy_to_user(arg, fi_args, sizeof(*fi_args)))
		ret = -EFAULT;
J
Jan Schmidt 已提交
2767

2768 2769
	kfree(fi_args);
	return ret;
J
Jan Schmidt 已提交
2770 2771
}

2772 2773
static long btrfs_ioctl_dev_info(struct btrfs_fs_info *fs_info,
				 void __user *arg)
J
Jan Schmidt 已提交
2774 2775 2776
{
	struct btrfs_ioctl_dev_info_args *di_args;
	struct btrfs_device *dev;
2777
	struct btrfs_fs_devices *fs_devices = fs_info->fs_devices;
J
Jan Schmidt 已提交
2778 2779 2780 2781 2782 2783 2784
	int ret = 0;
	char *s_uuid = NULL;

	di_args = memdup_user(arg, sizeof(*di_args));
	if (IS_ERR(di_args))
		return PTR_ERR(di_args);

2785
	if (!btrfs_is_empty_uuid(di_args->uuid))
J
Jan Schmidt 已提交
2786 2787 2788
		s_uuid = di_args->uuid;

	mutex_lock(&fs_devices->device_list_mutex);
2789
	dev = btrfs_find_device(fs_info, di_args->devid, s_uuid, NULL);
J
Jan Schmidt 已提交
2790 2791 2792 2793 2794 2795 2796

	if (!dev) {
		ret = -ENODEV;
		goto out;
	}

	di_args->devid = dev->devid;
2797 2798
	di_args->bytes_used = btrfs_device_get_bytes_used(dev);
	di_args->total_bytes = btrfs_device_get_total_bytes(dev);
J
Jan Schmidt 已提交
2799
	memcpy(di_args->uuid, dev->uuid, sizeof(di_args->uuid));
2800
	if (dev->name) {
2801 2802 2803 2804 2805 2806
		struct rcu_string *name;

		rcu_read_lock();
		name = rcu_dereference(dev->name);
		strncpy(di_args->path, name->str, sizeof(di_args->path));
		rcu_read_unlock();
2807 2808
		di_args->path[sizeof(di_args->path) - 1] = 0;
	} else {
2809
		di_args->path[0] = '\0';
2810
	}
J
Jan Schmidt 已提交
2811 2812

out:
2813
	mutex_unlock(&fs_devices->device_list_mutex);
J
Jan Schmidt 已提交
2814 2815 2816 2817 2818 2819 2820
	if (ret == 0 && copy_to_user(arg, di_args, sizeof(*di_args)))
		ret = -EFAULT;

	kfree(di_args);
	return ret;
}

2821
static struct page *extent_same_get_page(struct inode *inode, pgoff_t index)
M
Mark Fasheh 已提交
2822 2823 2824 2825 2826
{
	struct page *page;

	page = grab_cache_page(inode->i_mapping, index);
	if (!page)
2827
		return ERR_PTR(-ENOMEM);
M
Mark Fasheh 已提交
2828 2829

	if (!PageUptodate(page)) {
2830 2831 2832 2833 2834
		int ret;

		ret = btrfs_readpage(NULL, page);
		if (ret)
			return ERR_PTR(ret);
M
Mark Fasheh 已提交
2835 2836 2837
		lock_page(page);
		if (!PageUptodate(page)) {
			unlock_page(page);
2838
			put_page(page);
2839 2840 2841 2842
			return ERR_PTR(-EIO);
		}
		if (page->mapping != inode->i_mapping) {
			unlock_page(page);
2843
			put_page(page);
2844
			return ERR_PTR(-EAGAIN);
M
Mark Fasheh 已提交
2845 2846 2847 2848 2849 2850
		}
	}

	return page;
}

2851 2852 2853 2854
static int gather_extent_pages(struct inode *inode, struct page **pages,
			       int num_pages, u64 off)
{
	int i;
2855
	pgoff_t index = off >> PAGE_SHIFT;
2856 2857

	for (i = 0; i < num_pages; i++) {
2858
again:
2859
		pages[i] = extent_same_get_page(inode, index + i);
2860 2861 2862 2863 2864 2865 2866 2867
		if (IS_ERR(pages[i])) {
			int err = PTR_ERR(pages[i]);

			if (err == -EAGAIN)
				goto again;
			pages[i] = NULL;
			return err;
		}
2868 2869 2870 2871
	}
	return 0;
}

2872 2873
static int lock_extent_range(struct inode *inode, u64 off, u64 len,
			     bool retry_range_locking)
2874
{
2875 2876 2877 2878 2879 2880 2881 2882
	/*
	 * Do any pending delalloc/csum calculations on inode, one way or
	 * another, and lock file content.
	 * The locking order is:
	 *
	 *   1) pages
	 *   2) range in the inode's io tree
	 */
2883 2884 2885 2886 2887
	while (1) {
		struct btrfs_ordered_extent *ordered;
		lock_extent(&BTRFS_I(inode)->io_tree, off, off + len - 1);
		ordered = btrfs_lookup_first_ordered_extent(inode,
							    off + len - 1);
2888 2889 2890
		if ((!ordered ||
		     ordered->file_offset + ordered->len <= off ||
		     ordered->file_offset >= off + len) &&
2891
		    !test_range_bit(&BTRFS_I(inode)->io_tree, off,
2892 2893 2894
				    off + len - 1, EXTENT_DELALLOC, 0, NULL)) {
			if (ordered)
				btrfs_put_ordered_extent(ordered);
2895
			break;
2896
		}
2897 2898 2899
		unlock_extent(&BTRFS_I(inode)->io_tree, off, off + len - 1);
		if (ordered)
			btrfs_put_ordered_extent(ordered);
2900 2901
		if (!retry_range_locking)
			return -EAGAIN;
2902 2903
		btrfs_wait_ordered_range(inode, off, len);
	}
2904
	return 0;
2905 2906
}

2907
static void btrfs_double_inode_unlock(struct inode *inode1, struct inode *inode2)
M
Mark Fasheh 已提交
2908
{
A
Al Viro 已提交
2909 2910
	inode_unlock(inode1);
	inode_unlock(inode2);
M
Mark Fasheh 已提交
2911 2912
}

2913 2914 2915 2916 2917
static void btrfs_double_inode_lock(struct inode *inode1, struct inode *inode2)
{
	if (inode1 < inode2)
		swap(inode1, inode2);

A
Al Viro 已提交
2918 2919
	inode_lock_nested(inode1, I_MUTEX_PARENT);
	inode_lock_nested(inode2, I_MUTEX_CHILD);
2920 2921 2922 2923 2924 2925 2926 2927 2928
}

static void btrfs_double_extent_unlock(struct inode *inode1, u64 loff1,
				      struct inode *inode2, u64 loff2, u64 len)
{
	unlock_extent(&BTRFS_I(inode1)->io_tree, loff1, loff1 + len - 1);
	unlock_extent(&BTRFS_I(inode2)->io_tree, loff2, loff2 + len - 1);
}

2929 2930 2931
static int btrfs_double_extent_lock(struct inode *inode1, u64 loff1,
				    struct inode *inode2, u64 loff2, u64 len,
				    bool retry_range_locking)
M
Mark Fasheh 已提交
2932
{
2933 2934
	int ret;

M
Mark Fasheh 已提交
2935 2936 2937 2938
	if (inode1 < inode2) {
		swap(inode1, inode2);
		swap(loff1, loff2);
	}
2939 2940 2941 2942 2943 2944 2945 2946
	ret = lock_extent_range(inode1, loff1, len, retry_range_locking);
	if (ret)
		return ret;
	ret = lock_extent_range(inode2, loff2, len, retry_range_locking);
	if (ret)
		unlock_extent(&BTRFS_I(inode1)->io_tree, loff1,
			      loff1 + len - 1);
	return ret;
2947 2948 2949 2950 2951 2952 2953 2954 2955 2956 2957 2958 2959 2960 2961
}

struct cmp_pages {
	int		num_pages;
	struct page	**src_pages;
	struct page	**dst_pages;
};

static void btrfs_cmp_data_free(struct cmp_pages *cmp)
{
	int i;
	struct page *pg;

	for (i = 0; i < cmp->num_pages; i++) {
		pg = cmp->src_pages[i];
2962 2963
		if (pg) {
			unlock_page(pg);
2964
			put_page(pg);
2965
		}
2966
		pg = cmp->dst_pages[i];
2967 2968
		if (pg) {
			unlock_page(pg);
2969
			put_page(pg);
2970
		}
2971 2972 2973 2974 2975 2976 2977 2978 2979 2980
	}
	kfree(cmp->src_pages);
	kfree(cmp->dst_pages);
}

static int btrfs_cmp_data_prepare(struct inode *src, u64 loff,
				  struct inode *dst, u64 dst_loff,
				  u64 len, struct cmp_pages *cmp)
{
	int ret;
2981
	int num_pages = PAGE_ALIGN(len) >> PAGE_SHIFT;
2982 2983 2984 2985 2986 2987 2988 2989
	struct page **src_pgarr, **dst_pgarr;

	/*
	 * We must gather up all the pages before we initiate our
	 * extent locking. We use an array for the page pointers. Size
	 * of the array is bounded by len, which is in turn bounded by
	 * BTRFS_MAX_DEDUPE_LEN.
	 */
2990 2991
	src_pgarr = kcalloc(num_pages, sizeof(struct page *), GFP_KERNEL);
	dst_pgarr = kcalloc(num_pages, sizeof(struct page *), GFP_KERNEL);
2992 2993 2994 2995
	if (!src_pgarr || !dst_pgarr) {
		kfree(src_pgarr);
		kfree(dst_pgarr);
		return -ENOMEM;
M
Mark Fasheh 已提交
2996
	}
2997 2998 2999 3000
	cmp->num_pages = num_pages;
	cmp->src_pages = src_pgarr;
	cmp->dst_pages = dst_pgarr;

3001 3002 3003 3004 3005 3006 3007 3008 3009 3010 3011
	/*
	 * If deduping ranges in the same inode, locking rules make it mandatory
	 * to always lock pages in ascending order to avoid deadlocks with
	 * concurrent tasks (such as starting writeback/delalloc).
	 */
	if (src == dst && dst_loff < loff) {
		swap(src_pgarr, dst_pgarr);
		swap(loff, dst_loff);
	}

	ret = gather_extent_pages(src, src_pgarr, cmp->num_pages, loff);
3012 3013 3014
	if (ret)
		goto out;

3015
	ret = gather_extent_pages(dst, dst_pgarr, cmp->num_pages, dst_loff);
3016 3017 3018 3019

out:
	if (ret)
		btrfs_cmp_data_free(cmp);
3020
	return ret;
M
Mark Fasheh 已提交
3021 3022
}

3023
static int btrfs_cmp_data(u64 len, struct cmp_pages *cmp)
M
Mark Fasheh 已提交
3024 3025
{
	int ret = 0;
3026
	int i;
M
Mark Fasheh 已提交
3027
	struct page *src_page, *dst_page;
3028
	unsigned int cmp_len = PAGE_SIZE;
M
Mark Fasheh 已提交
3029 3030
	void *addr, *dst_addr;

3031
	i = 0;
M
Mark Fasheh 已提交
3032
	while (len) {
3033
		if (len < PAGE_SIZE)
M
Mark Fasheh 已提交
3034 3035
			cmp_len = len;

3036 3037 3038 3039
		BUG_ON(i >= cmp->num_pages);

		src_page = cmp->src_pages[i];
		dst_page = cmp->dst_pages[i];
3040 3041
		ASSERT(PageLocked(src_page));
		ASSERT(PageLocked(dst_page));
3042

M
Mark Fasheh 已提交
3043 3044 3045 3046 3047 3048 3049
		addr = kmap_atomic(src_page);
		dst_addr = kmap_atomic(dst_page);

		flush_dcache_page(src_page);
		flush_dcache_page(dst_page);

		if (memcmp(addr, dst_addr, cmp_len))
3050
			ret = -EBADE;
M
Mark Fasheh 已提交
3051 3052 3053 3054 3055 3056 3057 3058

		kunmap_atomic(addr);
		kunmap_atomic(dst_addr);

		if (ret)
			break;

		len -= cmp_len;
3059
		i++;
M
Mark Fasheh 已提交
3060 3061 3062 3063 3064
	}

	return ret;
}

3065 3066
static int extent_same_check_offsets(struct inode *inode, u64 off, u64 *plen,
				     u64 olen)
M
Mark Fasheh 已提交
3067
{
3068
	u64 len = *plen;
M
Mark Fasheh 已提交
3069 3070
	u64 bs = BTRFS_I(inode)->root->fs_info->sb->s_blocksize;

3071
	if (off + olen > inode->i_size || off + olen < off)
M
Mark Fasheh 已提交
3072
		return -EINVAL;
3073 3074 3075 3076 3077

	/* if we extend to eof, continue to block boundary */
	if (off + len == inode->i_size)
		*plen = len = ALIGN(inode->i_size, bs) - off;

M
Mark Fasheh 已提交
3078 3079 3080 3081 3082 3083 3084
	/* Check that we are block aligned - btrfs_clone() requires this */
	if (!IS_ALIGNED(off, bs) || !IS_ALIGNED(off + len, bs))
		return -EINVAL;

	return 0;
}

3085
static int btrfs_extent_same(struct inode *src, u64 loff, u64 olen,
M
Mark Fasheh 已提交
3086 3087 3088
			     struct inode *dst, u64 dst_loff)
{
	int ret;
3089
	u64 len = olen;
3090
	struct cmp_pages cmp;
3091
	bool same_inode = (src == dst);
M
Mark Fasheh 已提交
3092 3093
	u64 same_lock_start = 0;
	u64 same_lock_len = 0;
M
Mark Fasheh 已提交
3094

3095 3096 3097
	if (len == 0)
		return 0;

3098
	if (same_inode)
A
Al Viro 已提交
3099
		inode_lock(src);
3100 3101
	else
		btrfs_double_inode_lock(src, dst);
M
Mark Fasheh 已提交
3102

3103 3104 3105
	ret = extent_same_check_offsets(src, loff, &len, olen);
	if (ret)
		goto out_unlock;
M
Mark Fasheh 已提交
3106

3107 3108 3109 3110 3111
	ret = extent_same_check_offsets(dst, dst_loff, &len, olen);
	if (ret)
		goto out_unlock;

	if (same_inode) {
M
Mark Fasheh 已提交
3112 3113 3114 3115 3116 3117 3118 3119 3120 3121 3122 3123 3124 3125 3126 3127 3128 3129 3130 3131 3132 3133 3134 3135 3136 3137 3138 3139
		/*
		 * Single inode case wants the same checks, except we
		 * don't want our length pushed out past i_size as
		 * comparing that data range makes no sense.
		 *
		 * extent_same_check_offsets() will do this for an
		 * unaligned length at i_size, so catch it here and
		 * reject the request.
		 *
		 * This effectively means we require aligned extents
		 * for the single-inode case, whereas the other cases
		 * allow an unaligned length so long as it ends at
		 * i_size.
		 */
		if (len != olen) {
			ret = -EINVAL;
			goto out_unlock;
		}

		/* Check for overlapping ranges */
		if (dst_loff + len > loff && dst_loff < loff + len) {
			ret = -EINVAL;
			goto out_unlock;
		}

		same_lock_start = min_t(u64, loff, dst_loff);
		same_lock_len = max_t(u64, loff, dst_loff) + len - same_lock_start;
	}
M
Mark Fasheh 已提交
3140 3141 3142 3143 3144 3145 3146 3147

	/* don't make the dst file partly checksummed */
	if ((BTRFS_I(src)->flags & BTRFS_INODE_NODATASUM) !=
	    (BTRFS_I(dst)->flags & BTRFS_INODE_NODATASUM)) {
		ret = -EINVAL;
		goto out_unlock;
	}

3148
again:
3149 3150 3151 3152
	ret = btrfs_cmp_data_prepare(src, loff, dst, dst_loff, olen, &cmp);
	if (ret)
		goto out_unlock;

M
Mark Fasheh 已提交
3153
	if (same_inode)
3154 3155
		ret = lock_extent_range(src, same_lock_start, same_lock_len,
					false);
M
Mark Fasheh 已提交
3156
	else
3157 3158 3159 3160 3161 3162 3163 3164 3165 3166 3167 3168 3169 3170 3171 3172 3173 3174 3175 3176 3177 3178 3179 3180 3181 3182 3183 3184 3185 3186 3187
		ret = btrfs_double_extent_lock(src, loff, dst, dst_loff, len,
					       false);
	/*
	 * If one of the inodes has dirty pages in the respective range or
	 * ordered extents, we need to flush dellaloc and wait for all ordered
	 * extents in the range. We must unlock the pages and the ranges in the
	 * io trees to avoid deadlocks when flushing delalloc (requires locking
	 * pages) and when waiting for ordered extents to complete (they require
	 * range locking).
	 */
	if (ret == -EAGAIN) {
		/*
		 * Ranges in the io trees already unlocked. Now unlock all
		 * pages before waiting for all IO to complete.
		 */
		btrfs_cmp_data_free(&cmp);
		if (same_inode) {
			btrfs_wait_ordered_range(src, same_lock_start,
						 same_lock_len);
		} else {
			btrfs_wait_ordered_range(src, loff, len);
			btrfs_wait_ordered_range(dst, dst_loff, len);
		}
		goto again;
	}
	ASSERT(ret == 0);
	if (WARN_ON(ret)) {
		/* ranges in the io trees already unlocked */
		btrfs_cmp_data_free(&cmp);
		return ret;
	}
3188

3189
	/* pass original length for comparison so we stay within i_size */
3190
	ret = btrfs_cmp_data(olen, &cmp);
M
Mark Fasheh 已提交
3191
	if (ret == 0)
3192
		ret = btrfs_clone(src, dst, loff, olen, len, dst_loff, 1);
M
Mark Fasheh 已提交
3193

M
Mark Fasheh 已提交
3194 3195 3196 3197 3198
	if (same_inode)
		unlock_extent(&BTRFS_I(src)->io_tree, same_lock_start,
			      same_lock_start + same_lock_len - 1);
	else
		btrfs_double_extent_unlock(src, loff, dst, dst_loff, len);
3199 3200

	btrfs_cmp_data_free(&cmp);
M
Mark Fasheh 已提交
3201
out_unlock:
M
Mark Fasheh 已提交
3202
	if (same_inode)
A
Al Viro 已提交
3203
		inode_unlock(src);
M
Mark Fasheh 已提交
3204 3205
	else
		btrfs_double_inode_unlock(src, dst);
M
Mark Fasheh 已提交
3206 3207 3208 3209

	return ret;
}

3210
#define BTRFS_MAX_DEDUPE_LEN	SZ_16M
M
Mark Fasheh 已提交
3211

3212 3213
ssize_t btrfs_dedupe_file_range(struct file *src_file, u64 loff, u64 olen,
				struct file *dst_file, u64 dst_loff)
M
Mark Fasheh 已提交
3214
{
3215 3216
	struct inode *src = file_inode(src_file);
	struct inode *dst = file_inode(dst_file);
M
Mark Fasheh 已提交
3217
	u64 bs = BTRFS_I(src)->root->fs_info->sb->s_blocksize;
3218
	ssize_t res;
M
Mark Fasheh 已提交
3219

3220 3221
	if (olen > BTRFS_MAX_DEDUPE_LEN)
		olen = BTRFS_MAX_DEDUPE_LEN;
M
Mark Fasheh 已提交
3222

3223
	if (WARN_ON_ONCE(bs < PAGE_SIZE)) {
M
Mark Fasheh 已提交
3224 3225 3226 3227 3228
		/*
		 * Btrfs does not support blocksize < page_size. As a
		 * result, btrfs_cmp_data() won't correctly handle
		 * this situation without an update.
		 */
3229
		return -EINVAL;
M
Mark Fasheh 已提交
3230 3231
	}

3232 3233 3234 3235
	res = btrfs_extent_same(src, loff, olen, dst, dst_loff);
	if (res)
		return res;
	return olen;
M
Mark Fasheh 已提交
3236 3237
}

3238 3239 3240 3241
static int clone_finish_inode_update(struct btrfs_trans_handle *trans,
				     struct inode *inode,
				     u64 endoff,
				     const u64 destoff,
3242 3243
				     const u64 olen,
				     int no_time_update)
3244 3245 3246 3247 3248
{
	struct btrfs_root *root = BTRFS_I(inode)->root;
	int ret;

	inode_inc_iversion(inode);
3249
	if (!no_time_update)
3250
		inode->i_mtime = inode->i_ctime = current_time(inode);
3251 3252 3253 3254 3255 3256 3257
	/*
	 * We round up to the block size at eof when determining which
	 * extents to clone above, but shouldn't round up the file size.
	 */
	if (endoff > destoff + olen)
		endoff = destoff + olen;
	if (endoff > inode->i_size)
3258
		btrfs_i_size_write(BTRFS_I(inode), endoff);
3259 3260 3261

	ret = btrfs_update_inode(trans, root, inode);
	if (ret) {
3262
		btrfs_abort_transaction(trans, ret);
3263
		btrfs_end_transaction(trans);
3264 3265
		goto out;
	}
3266
	ret = btrfs_end_transaction(trans);
3267 3268 3269 3270
out:
	return ret;
}

3271
static void clone_update_extent_map(struct btrfs_inode *inode,
3272 3273 3274 3275 3276
				    const struct btrfs_trans_handle *trans,
				    const struct btrfs_path *path,
				    const u64 hole_offset,
				    const u64 hole_len)
{
3277
	struct extent_map_tree *em_tree = &inode->extent_tree;
3278 3279 3280 3281 3282
	struct extent_map *em;
	int ret;

	em = alloc_extent_map();
	if (!em) {
3283
		set_bit(BTRFS_INODE_NEEDS_FULL_SYNC, &inode->runtime_flags);
3284 3285 3286
		return;
	}

3287 3288 3289 3290 3291
	if (path) {
		struct btrfs_file_extent_item *fi;

		fi = btrfs_item_ptr(path->nodes[0], path->slots[0],
				    struct btrfs_file_extent_item);
3292
		btrfs_extent_item_to_extent_map(inode, path, fi, false, em);
3293 3294 3295 3296
		em->generation = -1;
		if (btrfs_file_extent_type(path->nodes[0], fi) ==
		    BTRFS_FILE_EXTENT_INLINE)
			set_bit(BTRFS_INODE_NEEDS_FULL_SYNC,
3297
					&inode->runtime_flags);
3298 3299 3300 3301 3302 3303 3304 3305 3306 3307 3308 3309 3310 3311 3312 3313 3314 3315 3316 3317
	} else {
		em->start = hole_offset;
		em->len = hole_len;
		em->ram_bytes = em->len;
		em->orig_start = hole_offset;
		em->block_start = EXTENT_MAP_HOLE;
		em->block_len = 0;
		em->orig_block_len = 0;
		em->compress_type = BTRFS_COMPRESS_NONE;
		em->generation = trans->transid;
	}

	while (1) {
		write_lock(&em_tree->lock);
		ret = add_extent_mapping(em_tree, em, 1);
		write_unlock(&em_tree->lock);
		if (ret != -EEXIST) {
			free_extent_map(em);
			break;
		}
3318
		btrfs_drop_extent_cache(inode, em->start,
3319 3320 3321
					em->start + em->len - 1, 0);
	}

3322
	if (ret)
3323
		set_bit(BTRFS_INODE_NEEDS_FULL_SYNC, &inode->runtime_flags);
3324 3325
}

3326 3327 3328 3329 3330 3331 3332 3333 3334 3335 3336 3337 3338 3339 3340 3341 3342 3343 3344 3345 3346 3347 3348 3349 3350
/*
 * Make sure we do not end up inserting an inline extent into a file that has
 * already other (non-inline) extents. If a file has an inline extent it can
 * not have any other extents and the (single) inline extent must start at the
 * file offset 0. Failing to respect these rules will lead to file corruption,
 * resulting in EIO errors on read/write operations, hitting BUG_ON's in mm, etc
 *
 * We can have extents that have been already written to disk or we can have
 * dirty ranges still in delalloc, in which case the extent maps and items are
 * created only when we run delalloc, and the delalloc ranges might fall outside
 * the range we are currently locking in the inode's io tree. So we check the
 * inode's i_size because of that (i_size updates are done while holding the
 * i_mutex, which we are holding here).
 * We also check to see if the inode has a size not greater than "datal" but has
 * extents beyond it, due to an fallocate with FALLOC_FL_KEEP_SIZE (and we are
 * protected against such concurrent fallocate calls by the i_mutex).
 *
 * If the file has no extents but a size greater than datal, do not allow the
 * copy because we would need turn the inline extent into a non-inline one (even
 * with NO_HOLES enabled). If we find our destination inode only has one inline
 * extent, just overwrite it with the source inline extent if its size is less
 * than the source extent's size, or we could copy the source inline extent's
 * data into the destination inode's inline extent if the later is greater then
 * the former.
 */
3351
static int clone_copy_inline_extent(struct inode *dst,
3352 3353 3354 3355 3356 3357 3358 3359 3360
				    struct btrfs_trans_handle *trans,
				    struct btrfs_path *path,
				    struct btrfs_key *new_key,
				    const u64 drop_start,
				    const u64 datal,
				    const u64 skip,
				    const u64 size,
				    char *inline_data)
{
3361
	struct btrfs_fs_info *fs_info = btrfs_sb(dst->i_sb);
3362 3363
	struct btrfs_root *root = BTRFS_I(dst)->root;
	const u64 aligned_end = ALIGN(new_key->offset + datal,
3364
				      fs_info->sectorsize);
3365 3366 3367 3368 3369 3370
	int ret;
	struct btrfs_key key;

	if (new_key->offset > 0)
		return -EOPNOTSUPP;

3371
	key.objectid = btrfs_ino(BTRFS_I(dst));
3372 3373 3374 3375 3376 3377 3378 3379 3380 3381 3382 3383 3384 3385
	key.type = BTRFS_EXTENT_DATA_KEY;
	key.offset = 0;
	ret = btrfs_search_slot(NULL, root, &key, path, 0, 0);
	if (ret < 0) {
		return ret;
	} else if (ret > 0) {
		if (path->slots[0] >= btrfs_header_nritems(path->nodes[0])) {
			ret = btrfs_next_leaf(root, path);
			if (ret < 0)
				return ret;
			else if (ret > 0)
				goto copy_inline_extent;
		}
		btrfs_item_key_to_cpu(path->nodes[0], &key, path->slots[0]);
3386
		if (key.objectid == btrfs_ino(BTRFS_I(dst)) &&
3387 3388 3389 3390 3391 3392 3393 3394 3395 3396 3397 3398 3399 3400 3401 3402 3403 3404 3405 3406 3407 3408 3409 3410 3411 3412 3413 3414 3415 3416 3417 3418 3419
		    key.type == BTRFS_EXTENT_DATA_KEY) {
			ASSERT(key.offset > 0);
			return -EOPNOTSUPP;
		}
	} else if (i_size_read(dst) <= datal) {
		struct btrfs_file_extent_item *ei;
		u64 ext_len;

		/*
		 * If the file size is <= datal, make sure there are no other
		 * extents following (can happen do to an fallocate call with
		 * the flag FALLOC_FL_KEEP_SIZE).
		 */
		ei = btrfs_item_ptr(path->nodes[0], path->slots[0],
				    struct btrfs_file_extent_item);
		/*
		 * If it's an inline extent, it can not have other extents
		 * following it.
		 */
		if (btrfs_file_extent_type(path->nodes[0], ei) ==
		    BTRFS_FILE_EXTENT_INLINE)
			goto copy_inline_extent;

		ext_len = btrfs_file_extent_num_bytes(path->nodes[0], ei);
		if (ext_len > aligned_end)
			return -EOPNOTSUPP;

		ret = btrfs_next_item(root, path);
		if (ret < 0) {
			return ret;
		} else if (ret == 0) {
			btrfs_item_key_to_cpu(path->nodes[0], &key,
					      path->slots[0]);
3420
			if (key.objectid == btrfs_ino(BTRFS_I(dst)) &&
3421 3422 3423 3424 3425 3426 3427 3428 3429 3430 3431 3432 3433 3434 3435 3436 3437 3438 3439 3440 3441 3442 3443 3444 3445 3446 3447 3448 3449 3450 3451 3452 3453 3454 3455 3456 3457 3458 3459 3460 3461 3462 3463 3464 3465 3466 3467 3468 3469
			    key.type == BTRFS_EXTENT_DATA_KEY)
				return -EOPNOTSUPP;
		}
	}

copy_inline_extent:
	/*
	 * We have no extent items, or we have an extent at offset 0 which may
	 * or may not be inlined. All these cases are dealt the same way.
	 */
	if (i_size_read(dst) > datal) {
		/*
		 * If the destination inode has an inline extent...
		 * This would require copying the data from the source inline
		 * extent into the beginning of the destination's inline extent.
		 * But this is really complex, both extents can be compressed
		 * or just one of them, which would require decompressing and
		 * re-compressing data (which could increase the new compressed
		 * size, not allowing the compressed data to fit anymore in an
		 * inline extent).
		 * So just don't support this case for now (it should be rare,
		 * we are not really saving space when cloning inline extents).
		 */
		return -EOPNOTSUPP;
	}

	btrfs_release_path(path);
	ret = btrfs_drop_extents(trans, root, dst, drop_start, aligned_end, 1);
	if (ret)
		return ret;
	ret = btrfs_insert_empty_item(trans, root, path, new_key, size);
	if (ret)
		return ret;

	if (skip) {
		const u32 start = btrfs_file_extent_calc_inline_size(0);

		memmove(inline_data + start, inline_data + start + skip, datal);
	}

	write_extent_buffer(path->nodes[0], inline_data,
			    btrfs_item_ptr_offset(path->nodes[0],
						  path->slots[0]),
			    size);
	inode_add_bytes(dst, datal);

	return 0;
}

3470 3471 3472 3473 3474 3475 3476
/**
 * btrfs_clone() - clone a range from inode file to another
 *
 * @src: Inode to clone from
 * @inode: Inode to clone to
 * @off: Offset within source to start clone from
 * @olen: Original length, passed by user, of range to clone
3477
 * @olen_aligned: Block-aligned value of olen
3478
 * @destoff: Offset within @inode to start clone
3479
 * @no_time_update: Whether to update mtime/ctime on the target inode
3480 3481
 */
static int btrfs_clone(struct inode *src, struct inode *inode,
3482
		       const u64 off, const u64 olen, const u64 olen_aligned,
3483
		       const u64 destoff, int no_time_update)
C
Christoph Hellwig 已提交
3484
{
3485
	struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb);
C
Christoph Hellwig 已提交
3486
	struct btrfs_root *root = BTRFS_I(inode)->root;
3487
	struct btrfs_path *path = NULL;
C
Christoph Hellwig 已提交
3488
	struct extent_buffer *leaf;
3489 3490
	struct btrfs_trans_handle *trans;
	char *buf = NULL;
Y
Yan Zheng 已提交
3491
	struct btrfs_key key;
C
Christoph Hellwig 已提交
3492 3493
	u32 nritems;
	int slot;
Y
Yan Zheng 已提交
3494
	int ret;
3495 3496
	const u64 len = olen_aligned;
	u64 last_dest_end = destoff;
Y
Yan Zheng 已提交
3497 3498

	ret = -ENOMEM;
3499 3500 3501
	buf = kvmalloc(fs_info->nodesize, GFP_KERNEL);
	if (!buf)
		return ret;
Y
Yan Zheng 已提交
3502 3503 3504

	path = btrfs_alloc_path();
	if (!path) {
3505
		kvfree(buf);
3506
		return ret;
3507 3508
	}

3509
	path->reada = READA_FORWARD;
3510
	/* clone data */
3511
	key.objectid = btrfs_ino(BTRFS_I(src));
Y
Yan Zheng 已提交
3512
	key.type = BTRFS_EXTENT_DATA_KEY;
3513
	key.offset = off;
C
Christoph Hellwig 已提交
3514 3515

	while (1) {
3516
		u64 next_key_min_offset = key.offset + 1;
3517

C
Christoph Hellwig 已提交
3518 3519 3520 3521
		/*
		 * note the key will change type as we walk through the
		 * tree.
		 */
3522
		path->leave_spinning = 1;
3523 3524
		ret = btrfs_search_slot(NULL, BTRFS_I(src)->root, &key, path,
				0, 0);
C
Christoph Hellwig 已提交
3525 3526
		if (ret < 0)
			goto out;
3527 3528 3529 3530 3531 3532 3533 3534 3535 3536 3537
		/*
		 * First search, if no extent item that starts at offset off was
		 * found but the previous item is an extent item, it's possible
		 * it might overlap our target range, therefore process it.
		 */
		if (key.offset == off && ret > 0 && path->slots[0] > 0) {
			btrfs_item_key_to_cpu(path->nodes[0], &key,
					      path->slots[0] - 1);
			if (key.type == BTRFS_EXTENT_DATA_KEY)
				path->slots[0]--;
		}
C
Christoph Hellwig 已提交
3538

Y
Yan Zheng 已提交
3539
		nritems = btrfs_header_nritems(path->nodes[0]);
3540
process_slot:
Y
Yan Zheng 已提交
3541
		if (path->slots[0] >= nritems) {
3542
			ret = btrfs_next_leaf(BTRFS_I(src)->root, path);
C
Christoph Hellwig 已提交
3543 3544 3545 3546
			if (ret < 0)
				goto out;
			if (ret > 0)
				break;
Y
Yan Zheng 已提交
3547
			nritems = btrfs_header_nritems(path->nodes[0]);
C
Christoph Hellwig 已提交
3548 3549 3550 3551
		}
		leaf = path->nodes[0];
		slot = path->slots[0];

Y
Yan Zheng 已提交
3552
		btrfs_item_key_to_cpu(leaf, &key, slot);
3553
		if (key.type > BTRFS_EXTENT_DATA_KEY ||
3554
		    key.objectid != btrfs_ino(BTRFS_I(src)))
C
Christoph Hellwig 已提交
3555 3556
			break;

3557
		if (key.type == BTRFS_EXTENT_DATA_KEY) {
3558 3559
			struct btrfs_file_extent_item *extent;
			int type;
Z
Zheng Yan 已提交
3560 3561
			u32 size;
			struct btrfs_key new_key;
3562 3563 3564
			u64 disko = 0, diskl = 0;
			u64 datao = 0, datal = 0;
			u8 comp;
3565
			u64 drop_start;
Z
Zheng Yan 已提交
3566

3567 3568 3569 3570
			extent = btrfs_item_ptr(leaf, slot,
						struct btrfs_file_extent_item);
			comp = btrfs_file_extent_compression(leaf, extent);
			type = btrfs_file_extent_type(leaf, extent);
3571 3572
			if (type == BTRFS_FILE_EXTENT_REG ||
			    type == BTRFS_FILE_EXTENT_PREALLOC) {
C
Chris Mason 已提交
3573 3574 3575 3576
				disko = btrfs_file_extent_disk_bytenr(leaf,
								      extent);
				diskl = btrfs_file_extent_disk_num_bytes(leaf,
								 extent);
3577
				datao = btrfs_file_extent_offset(leaf, extent);
C
Chris Mason 已提交
3578 3579
				datal = btrfs_file_extent_num_bytes(leaf,
								    extent);
3580 3581 3582 3583 3584
			} else if (type == BTRFS_FILE_EXTENT_INLINE) {
				/* take upper bound, may be compressed */
				datal = btrfs_file_extent_ram_bytes(leaf,
								    extent);
			}
Z
Zheng Yan 已提交
3585

3586 3587 3588 3589 3590 3591
			/*
			 * The first search might have left us at an extent
			 * item that ends before our target range's start, can
			 * happen if we have holes and NO_HOLES feature enabled.
			 */
			if (key.offset + datal <= off) {
3592 3593
				path->slots[0]++;
				goto process_slot;
3594 3595
			} else if (key.offset >= off + len) {
				break;
3596
			}
3597
			next_key_min_offset = key.offset + datal;
3598 3599 3600 3601 3602 3603 3604
			size = btrfs_item_size_nr(leaf, slot);
			read_extent_buffer(leaf, buf,
					   btrfs_item_ptr_offset(leaf, slot),
					   size);

			btrfs_release_path(path);
			path->leave_spinning = 0;
3605

Z
Zheng Yan 已提交
3606
			memcpy(&new_key, &key, sizeof(new_key));
3607
			new_key.objectid = btrfs_ino(BTRFS_I(inode));
3608 3609 3610 3611
			if (off <= key.offset)
				new_key.offset = key.offset + destoff - off;
			else
				new_key.offset = destoff;
Z
Zheng Yan 已提交
3612

3613 3614 3615 3616 3617 3618 3619 3620 3621 3622 3623 3624
			/*
			 * Deal with a hole that doesn't have an extent item
			 * that represents it (NO_HOLES feature enabled).
			 * This hole is either in the middle of the cloning
			 * range or at the beginning (fully overlaps it or
			 * partially overlaps it).
			 */
			if (new_key.offset != last_dest_end)
				drop_start = last_dest_end;
			else
				drop_start = new_key.offset;

3625 3626 3627 3628 3629 3630
			/*
			 * 1 - adjusting old extent (we may have to split it)
			 * 1 - add new extent
			 * 1 - inode update
			 */
			trans = btrfs_start_transaction(root, 3);
3631 3632 3633 3634 3635
			if (IS_ERR(trans)) {
				ret = PTR_ERR(trans);
				goto out;
			}

3636 3637
			if (type == BTRFS_FILE_EXTENT_REG ||
			    type == BTRFS_FILE_EXTENT_PREALLOC) {
3638 3639 3640 3641 3642
				/*
				 *    a  | --- range to clone ---|  b
				 * | ------------- extent ------------- |
				 */

3643
				/* subtract range b */
3644 3645 3646
				if (key.offset + datal > off + len)
					datal = off + len - key.offset;

3647
				/* subtract range a */
3648 3649 3650 3651 3652
				if (off > key.offset) {
					datao += off - key.offset;
					datal -= off - key.offset;
				}

J
Josef Bacik 已提交
3653
				ret = btrfs_drop_extents(trans, root, inode,
3654
							 drop_start,
3655
							 new_key.offset + datal,
3656
							 1);
3657
				if (ret) {
3658
					if (ret != -EOPNOTSUPP)
3659
						btrfs_abort_transaction(trans,
3660
									ret);
3661
					btrfs_end_transaction(trans);
3662 3663
					goto out;
				}
3664

3665 3666
				ret = btrfs_insert_empty_item(trans, root, path,
							      &new_key, size);
3667
				if (ret) {
3668
					btrfs_abort_transaction(trans, ret);
3669
					btrfs_end_transaction(trans);
3670 3671
					goto out;
				}
3672 3673 3674 3675

				leaf = path->nodes[0];
				slot = path->slots[0];
				write_extent_buffer(leaf, buf,
Z
Zheng Yan 已提交
3676 3677
					    btrfs_item_ptr_offset(leaf, slot),
					    size);
Y
Yan Zheng 已提交
3678

3679
				extent = btrfs_item_ptr(leaf, slot,
C
Christoph Hellwig 已提交
3680
						struct btrfs_file_extent_item);
3681 3682 3683 3684 3685 3686 3687 3688 3689

				/* disko == 0 means it's a hole */
				if (!disko)
					datao = 0;

				btrfs_set_file_extent_offset(leaf, extent,
							     datao);
				btrfs_set_file_extent_num_bytes(leaf, extent,
								datal);
J
Josef Bacik 已提交
3690

3691 3692
				if (disko) {
					inode_add_bytes(inode, datal);
3693 3694
					ret = btrfs_inc_extent_ref(trans,
							fs_info,
3695 3696
							disko, diskl, 0,
							root->root_key.objectid,
3697
							btrfs_ino(BTRFS_I(inode)),
3698
							new_key.offset - datao);
3699 3700 3701
					if (ret) {
						btrfs_abort_transaction(trans,
									ret);
3702
						btrfs_end_transaction(trans);
3703 3704 3705
						goto out;

					}
C
Christoph Hellwig 已提交
3706
				}
3707 3708 3709
			} else if (type == BTRFS_FILE_EXTENT_INLINE) {
				u64 skip = 0;
				u64 trim = 0;
3710

3711 3712 3713 3714
				if (off > key.offset) {
					skip = off - key.offset;
					new_key.offset += skip;
				}
C
Chris Mason 已提交
3715

L
Liu Bo 已提交
3716 3717
				if (key.offset + datal > off + len)
					trim = key.offset + datal - (off + len);
C
Chris Mason 已提交
3718

3719 3720
				if (comp && (skip || trim)) {
					ret = -EINVAL;
3721
					btrfs_end_transaction(trans);
3722 3723 3724 3725
					goto out;
				}
				size -= skip + trim;
				datal -= skip + trim;
3726

3727
				ret = clone_copy_inline_extent(inode,
3728 3729 3730 3731 3732
							       trans, path,
							       &new_key,
							       drop_start,
							       datal,
							       skip, size, buf);
3733
				if (ret) {
3734
					if (ret != -EOPNOTSUPP)
3735
						btrfs_abort_transaction(trans,
3736
									ret);
3737
					btrfs_end_transaction(trans);
3738 3739
					goto out;
				}
3740 3741
				leaf = path->nodes[0];
				slot = path->slots[0];
C
Christoph Hellwig 已提交
3742
			}
3743

3744 3745
			/* If we have an implicit hole (NO_HOLES feature). */
			if (drop_start < new_key.offset)
3746
				clone_update_extent_map(BTRFS_I(inode), trans,
3747
						NULL, drop_start,
3748 3749
						new_key.offset - drop_start);

3750 3751
			clone_update_extent_map(BTRFS_I(inode), trans,
					path, 0, 0);
3752

3753
			btrfs_mark_buffer_dirty(leaf);
3754
			btrfs_release_path(path);
3755

3756
			last_dest_end = ALIGN(new_key.offset + datal,
3757
					      fs_info->sectorsize);
3758 3759
			ret = clone_finish_inode_update(trans, inode,
							last_dest_end,
3760 3761
							destoff, olen,
							no_time_update);
3762
			if (ret)
3763
				goto out;
3764 3765
			if (new_key.offset + datal >= destoff + len)
				break;
3766
		}
3767
		btrfs_release_path(path);
3768
		key.offset = next_key_min_offset;
3769 3770 3771 3772 3773

		if (fatal_signal_pending(current)) {
			ret = -EINTR;
			goto out;
		}
C
Christoph Hellwig 已提交
3774 3775
	}
	ret = 0;
3776

3777 3778 3779 3780 3781 3782 3783 3784 3785 3786 3787 3788 3789 3790 3791 3792 3793 3794 3795 3796
	if (last_dest_end < destoff + len) {
		/*
		 * We have an implicit hole (NO_HOLES feature is enabled) that
		 * fully or partially overlaps our cloning range at its end.
		 */
		btrfs_release_path(path);

		/*
		 * 1 - remove extent(s)
		 * 1 - inode update
		 */
		trans = btrfs_start_transaction(root, 2);
		if (IS_ERR(trans)) {
			ret = PTR_ERR(trans);
			goto out;
		}
		ret = btrfs_drop_extents(trans, root, inode,
					 last_dest_end, destoff + len, 1);
		if (ret) {
			if (ret != -EOPNOTSUPP)
3797
				btrfs_abort_transaction(trans, ret);
3798
			btrfs_end_transaction(trans);
3799 3800
			goto out;
		}
3801 3802 3803
		clone_update_extent_map(BTRFS_I(inode), trans, NULL,
				last_dest_end,
				destoff + len - last_dest_end);
3804
		ret = clone_finish_inode_update(trans, inode, destoff + len,
3805
						destoff, olen, no_time_update);
3806 3807
	}

C
Christoph Hellwig 已提交
3808
out:
3809
	btrfs_free_path(path);
3810
	kvfree(buf);
3811 3812 3813
	return ret;
}

3814 3815
static noinline int btrfs_clone_files(struct file *file, struct file *file_src,
					u64 off, u64 olen, u64 destoff)
3816
{
A
Al Viro 已提交
3817
	struct inode *inode = file_inode(file);
3818
	struct inode *src = file_inode(file_src);
3819
	struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb);
3820 3821 3822
	struct btrfs_root *root = BTRFS_I(inode)->root;
	int ret;
	u64 len = olen;
3823
	u64 bs = fs_info->sb->s_blocksize;
3824
	int same_inode = src == inode;
3825 3826 3827 3828 3829 3830 3831

	/*
	 * TODO:
	 * - split compressed inline extents.  annoying: we need to
	 *   decompress into destination's address_space (the file offset
	 *   may change, so source mapping won't do), then recompress (or
	 *   otherwise reinsert) a subrange.
3832 3833 3834
	 *
	 * - split destination inode's inline extents.  The inline extents can
	 *   be either compressed or non-compressed.
3835 3836 3837 3838 3839
	 */

	if (btrfs_root_readonly(root))
		return -EROFS;

3840 3841 3842
	if (file_src->f_path.mnt != file->f_path.mnt ||
	    src->i_sb != inode->i_sb)
		return -EXDEV;
3843 3844 3845 3846

	/* don't make the dst file partly checksummed */
	if ((BTRFS_I(src)->flags & BTRFS_INODE_NODATASUM) !=
	    (BTRFS_I(inode)->flags & BTRFS_INODE_NODATASUM))
3847
		return -EINVAL;
3848 3849

	if (S_ISDIR(src->i_mode) || S_ISDIR(inode->i_mode))
3850
		return -EISDIR;
3851 3852

	if (!same_inode) {
3853
		btrfs_double_inode_lock(src, inode);
3854
	} else {
A
Al Viro 已提交
3855
		inode_lock(src);
3856 3857 3858 3859 3860 3861 3862 3863 3864 3865 3866 3867
	}

	/* determine range to clone */
	ret = -EINVAL;
	if (off + len > src->i_size || off + len < off)
		goto out_unlock;
	if (len == 0)
		olen = len = src->i_size - off;
	/* if we extend to eof, continue to block boundary */
	if (off + len == src->i_size)
		len = ALIGN(src->i_size, bs) - off;

3868 3869 3870 3871 3872
	if (len == 0) {
		ret = 0;
		goto out_unlock;
	}

3873 3874 3875 3876 3877 3878 3879 3880 3881 3882 3883 3884 3885 3886 3887 3888 3889
	/* verify the end result is block aligned */
	if (!IS_ALIGNED(off, bs) || !IS_ALIGNED(off + len, bs) ||
	    !IS_ALIGNED(destoff, bs))
		goto out_unlock;

	/* verify if ranges are overlapped within the same file */
	if (same_inode) {
		if (destoff + len > off && destoff < off + len)
			goto out_unlock;
	}

	if (destoff > inode->i_size) {
		ret = btrfs_cont_expand(inode, inode->i_size, destoff);
		if (ret)
			goto out_unlock;
	}

3890 3891 3892 3893 3894 3895 3896 3897 3898 3899
	/*
	 * Lock the target range too. Right after we replace the file extent
	 * items in the fs tree (which now point to the cloned data), we might
	 * have a worker replace them with extent items relative to a write
	 * operation that was issued before this clone operation (i.e. confront
	 * with inode.c:btrfs_finish_ordered_io).
	 */
	if (same_inode) {
		u64 lock_start = min_t(u64, off, destoff);
		u64 lock_len = max_t(u64, off, destoff) + len - lock_start;
3900

3901
		ret = lock_extent_range(src, lock_start, lock_len, true);
3902
	} else {
3903 3904 3905 3906 3907 3908 3909
		ret = btrfs_double_extent_lock(src, off, inode, destoff, len,
					       true);
	}
	ASSERT(ret == 0);
	if (WARN_ON(ret)) {
		/* ranges in the io trees already unlocked */
		goto out_unlock;
3910
	}
3911

3912
	ret = btrfs_clone(src, inode, off, olen, len, destoff, 0);
3913

3914 3915 3916 3917 3918 3919
	if (same_inode) {
		u64 lock_start = min_t(u64, off, destoff);
		u64 lock_end = max_t(u64, off, destoff) + len - 1;

		unlock_extent(&BTRFS_I(src)->io_tree, lock_start, lock_end);
	} else {
3920
		btrfs_double_extent_unlock(src, off, inode, destoff, len);
3921 3922 3923 3924 3925
	}
	/*
	 * Truncate page cache pages so that future reads will see the cloned
	 * data immediately and not the previous data.
	 */
3926
	truncate_inode_pages_range(&inode->i_data,
3927 3928
				round_down(destoff, PAGE_SIZE),
				round_up(destoff + len, PAGE_SIZE) - 1);
C
Christoph Hellwig 已提交
3929
out_unlock:
3930 3931 3932
	if (!same_inode)
		btrfs_double_inode_unlock(src, inode);
	else
A
Al Viro 已提交
3933
		inode_unlock(src);
3934 3935 3936
	return ret;
}

3937 3938
int btrfs_clone_file_range(struct file *src_file, loff_t off,
		struct file *dst_file, loff_t destoff, u64 len)
3939
{
3940
	return btrfs_clone_files(dst_file, src_file, off, len, destoff);
3941 3942
}

C
Christoph Hellwig 已提交
3943 3944 3945 3946 3947 3948
/*
 * there are many ways the trans_start and trans_end ioctls can lead
 * to deadlocks.  They should only be used by applications that
 * basically own the machine, and have a very in depth understanding
 * of all the possible deadlocks and enospc problems.
 */
3949
static long btrfs_ioctl_trans_start(struct file *file)
C
Christoph Hellwig 已提交
3950
{
A
Al Viro 已提交
3951
	struct inode *inode = file_inode(file);
3952
	struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb);
C
Christoph Hellwig 已提交
3953 3954
	struct btrfs_root *root = BTRFS_I(inode)->root;
	struct btrfs_trans_handle *trans;
3955
	struct btrfs_file_private *private;
3956
	int ret;
3957
	static bool warned = false;
C
Christoph Hellwig 已提交
3958

3959
	ret = -EPERM;
3960
	if (!capable(CAP_SYS_ADMIN))
3961
		goto out;
3962

3963 3964 3965 3966 3967 3968 3969 3970 3971 3972
	if (!warned) {
		btrfs_warn(fs_info,
			"Userspace transaction mechanism is considered "
			"deprecated and slated to be removed in 4.17. "
			"If you have a valid use case please "
			"speak up on the mailing list");
		WARN_ON(1);
		warned = true;
	}

3973
	ret = -EINPROGRESS;
3974 3975
	private = file->private_data;
	if (private && private->trans)
C
Christoph Hellwig 已提交
3976
		goto out;
3977 3978 3979 3980 3981 3982 3983
	if (!private) {
		private = kzalloc(sizeof(struct btrfs_file_private),
				  GFP_KERNEL);
		if (!private)
			return -ENOMEM;
		file->private_data = private;
	}
3984

L
Li Zefan 已提交
3985 3986 3987 3988
	ret = -EROFS;
	if (btrfs_root_readonly(root))
		goto out;

3989
	ret = mnt_want_write_file(file);
Y
Yan Zheng 已提交
3990 3991 3992
	if (ret)
		goto out;

3993
	atomic_inc(&fs_info->open_ioctl_trans);
3994

3995
	ret = -ENOMEM;
3996
	trans = btrfs_start_ioctl_transaction(root);
3997
	if (IS_ERR(trans))
3998 3999
		goto out_drop;

4000
	private->trans = trans;
4001 4002 4003
	return 0;

out_drop:
4004
	atomic_dec(&fs_info->open_ioctl_trans);
A
Al Viro 已提交
4005
	mnt_drop_write_file(file);
C
Christoph Hellwig 已提交
4006 4007 4008 4009
out:
	return ret;
}

4010 4011
static long btrfs_ioctl_default_subvol(struct file *file, void __user *argp)
{
A
Al Viro 已提交
4012
	struct inode *inode = file_inode(file);
4013
	struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb);
4014 4015 4016 4017 4018 4019 4020 4021 4022
	struct btrfs_root *root = BTRFS_I(inode)->root;
	struct btrfs_root *new_root;
	struct btrfs_dir_item *di;
	struct btrfs_trans_handle *trans;
	struct btrfs_path *path;
	struct btrfs_key location;
	struct btrfs_disk_key disk_key;
	u64 objectid = 0;
	u64 dir_id;
4023
	int ret;
4024 4025 4026 4027

	if (!capable(CAP_SYS_ADMIN))
		return -EPERM;

4028 4029 4030 4031 4032 4033 4034 4035
	ret = mnt_want_write_file(file);
	if (ret)
		return ret;

	if (copy_from_user(&objectid, argp, sizeof(objectid))) {
		ret = -EFAULT;
		goto out;
	}
4036 4037

	if (!objectid)
4038
		objectid = BTRFS_FS_TREE_OBJECTID;
4039 4040 4041 4042 4043

	location.objectid = objectid;
	location.type = BTRFS_ROOT_ITEM_KEY;
	location.offset = (u64)-1;

4044
	new_root = btrfs_read_fs_root_no_name(fs_info, &location);
4045 4046 4047 4048
	if (IS_ERR(new_root)) {
		ret = PTR_ERR(new_root);
		goto out;
	}
4049 4050 4051 4052
	if (!is_fstree(new_root->objectid)) {
		ret = -ENOENT;
		goto out;
	}
4053 4054

	path = btrfs_alloc_path();
4055 4056 4057 4058
	if (!path) {
		ret = -ENOMEM;
		goto out;
	}
4059 4060 4061
	path->leave_spinning = 1;

	trans = btrfs_start_transaction(root, 1);
4062
	if (IS_ERR(trans)) {
4063
		btrfs_free_path(path);
4064 4065
		ret = PTR_ERR(trans);
		goto out;
4066 4067
	}

4068 4069
	dir_id = btrfs_super_root_dir(fs_info->super_copy);
	di = btrfs_lookup_dir_item(trans, fs_info->tree_root, path,
4070
				   dir_id, "default", 7, 1);
4071
	if (IS_ERR_OR_NULL(di)) {
4072
		btrfs_free_path(path);
4073
		btrfs_end_transaction(trans);
4074
		btrfs_err(fs_info,
J
Jeff Mahoney 已提交
4075
			  "Umm, you don't have the default diritem, this isn't going to work");
4076 4077
		ret = -ENOENT;
		goto out;
4078 4079 4080 4081 4082 4083 4084
	}

	btrfs_cpu_key_to_disk(&disk_key, &new_root->root_key);
	btrfs_set_dir_item_key(path->nodes[0], di, &disk_key);
	btrfs_mark_buffer_dirty(path->nodes[0]);
	btrfs_free_path(path);

4085
	btrfs_set_fs_incompat(fs_info, DEFAULT_SUBVOL);
4086
	btrfs_end_transaction(trans);
4087 4088 4089
out:
	mnt_drop_write_file(file);
	return ret;
4090 4091
}

4092 4093
void btrfs_get_block_group_info(struct list_head *groups_list,
				struct btrfs_ioctl_space_info *space)
4094 4095 4096 4097 4098 4099 4100 4101 4102 4103 4104 4105 4106 4107
{
	struct btrfs_block_group_cache *block_group;

	space->total_bytes = 0;
	space->used_bytes = 0;
	space->flags = 0;
	list_for_each_entry(block_group, groups_list, list) {
		space->flags = block_group->flags;
		space->total_bytes += block_group->key.offset;
		space->used_bytes +=
			btrfs_block_group_used(&block_group->item);
	}
}

4108 4109
static long btrfs_ioctl_space_info(struct btrfs_fs_info *fs_info,
				   void __user *arg)
J
Josef Bacik 已提交
4110 4111 4112 4113
{
	struct btrfs_ioctl_space_args space_args;
	struct btrfs_ioctl_space_info space;
	struct btrfs_ioctl_space_info *dest;
4114
	struct btrfs_ioctl_space_info *dest_orig;
4115
	struct btrfs_ioctl_space_info __user *user_dest;
J
Josef Bacik 已提交
4116
	struct btrfs_space_info *info;
4117 4118 4119 4120 4121
	u64 types[] = {BTRFS_BLOCK_GROUP_DATA,
		       BTRFS_BLOCK_GROUP_SYSTEM,
		       BTRFS_BLOCK_GROUP_METADATA,
		       BTRFS_BLOCK_GROUP_DATA | BTRFS_BLOCK_GROUP_METADATA};
	int num_types = 4;
4122
	int alloc_size;
J
Josef Bacik 已提交
4123
	int ret = 0;
4124
	u64 slot_count = 0;
4125
	int i, c;
J
Josef Bacik 已提交
4126 4127 4128 4129 4130 4131

	if (copy_from_user(&space_args,
			   (struct btrfs_ioctl_space_args __user *)arg,
			   sizeof(space_args)))
		return -EFAULT;

4132 4133 4134 4135 4136
	for (i = 0; i < num_types; i++) {
		struct btrfs_space_info *tmp;

		info = NULL;
		rcu_read_lock();
4137
		list_for_each_entry_rcu(tmp, &fs_info->space_info,
4138 4139 4140 4141 4142 4143 4144 4145 4146 4147 4148 4149 4150 4151 4152 4153 4154 4155
					list) {
			if (tmp->flags == types[i]) {
				info = tmp;
				break;
			}
		}
		rcu_read_unlock();

		if (!info)
			continue;

		down_read(&info->groups_sem);
		for (c = 0; c < BTRFS_NR_RAID_TYPES; c++) {
			if (!list_empty(&info->block_groups[c]))
				slot_count++;
		}
		up_read(&info->groups_sem);
	}
4156

4157 4158 4159 4160 4161
	/*
	 * Global block reserve, exported as a space_info
	 */
	slot_count++;

4162 4163 4164 4165 4166
	/* space_slots == 0 means they are asking for a count */
	if (space_args.space_slots == 0) {
		space_args.total_spaces = slot_count;
		goto out;
	}
4167

4168
	slot_count = min_t(u64, space_args.space_slots, slot_count);
4169

4170
	alloc_size = sizeof(*dest) * slot_count;
4171

4172 4173 4174
	/* we generally have at most 6 or so space infos, one for each raid
	 * level.  So, a whole page should be more than enough for everyone
	 */
4175
	if (alloc_size > PAGE_SIZE)
4176 4177
		return -ENOMEM;

J
Josef Bacik 已提交
4178
	space_args.total_spaces = 0;
4179
	dest = kmalloc(alloc_size, GFP_KERNEL);
4180 4181 4182
	if (!dest)
		return -ENOMEM;
	dest_orig = dest;
J
Josef Bacik 已提交
4183

4184
	/* now we have a buffer to copy into */
4185 4186 4187
	for (i = 0; i < num_types; i++) {
		struct btrfs_space_info *tmp;

4188 4189 4190
		if (!slot_count)
			break;

4191 4192
		info = NULL;
		rcu_read_lock();
4193
		list_for_each_entry_rcu(tmp, &fs_info->space_info,
4194 4195 4196 4197 4198 4199 4200
					list) {
			if (tmp->flags == types[i]) {
				info = tmp;
				break;
			}
		}
		rcu_read_unlock();
4201

4202 4203 4204 4205 4206
		if (!info)
			continue;
		down_read(&info->groups_sem);
		for (c = 0; c < BTRFS_NR_RAID_TYPES; c++) {
			if (!list_empty(&info->block_groups[c])) {
4207 4208
				btrfs_get_block_group_info(
					&info->block_groups[c], &space);
4209 4210 4211
				memcpy(dest, &space, sizeof(space));
				dest++;
				space_args.total_spaces++;
4212
				slot_count--;
4213
			}
4214 4215
			if (!slot_count)
				break;
4216 4217
		}
		up_read(&info->groups_sem);
J
Josef Bacik 已提交
4218 4219
	}

4220 4221 4222 4223
	/*
	 * Add global block reserve
	 */
	if (slot_count) {
4224
		struct btrfs_block_rsv *block_rsv = &fs_info->global_block_rsv;
4225 4226 4227 4228 4229 4230 4231 4232 4233 4234

		spin_lock(&block_rsv->lock);
		space.total_bytes = block_rsv->size;
		space.used_bytes = block_rsv->size - block_rsv->reserved;
		spin_unlock(&block_rsv->lock);
		space.flags = BTRFS_SPACE_INFO_GLOBAL_RSV;
		memcpy(dest, &space, sizeof(space));
		space_args.total_spaces++;
	}

D
Daniel J Blueman 已提交
4235
	user_dest = (struct btrfs_ioctl_space_info __user *)
4236 4237 4238 4239 4240 4241 4242 4243
		(arg + sizeof(struct btrfs_ioctl_space_args));

	if (copy_to_user(user_dest, dest_orig, alloc_size))
		ret = -EFAULT;

	kfree(dest_orig);
out:
	if (ret == 0 && copy_to_user(arg, &space_args, sizeof(space_args)))
J
Josef Bacik 已提交
4244 4245 4246 4247 4248
		ret = -EFAULT;

	return ret;
}

C
Christoph Hellwig 已提交
4249 4250 4251 4252 4253 4254 4255 4256
/*
 * there are many ways the trans_start and trans_end ioctls can lead
 * to deadlocks.  They should only be used by applications that
 * basically own the machine, and have a very in depth understanding
 * of all the possible deadlocks and enospc problems.
 */
long btrfs_ioctl_trans_end(struct file *file)
{
A
Al Viro 已提交
4257
	struct inode *inode = file_inode(file);
C
Christoph Hellwig 已提交
4258
	struct btrfs_root *root = BTRFS_I(inode)->root;
4259
	struct btrfs_file_private *private = file->private_data;
C
Christoph Hellwig 已提交
4260

4261
	if (!private || !private->trans)
4262
		return -EINVAL;
4263

4264 4265
	btrfs_end_transaction(private->trans);
	private->trans = NULL;
4266

J
Josef Bacik 已提交
4267
	atomic_dec(&root->fs_info->open_ioctl_trans);
4268

A
Al Viro 已提交
4269
	mnt_drop_write_file(file);
4270
	return 0;
C
Christoph Hellwig 已提交
4271 4272
}

4273 4274
static noinline long btrfs_ioctl_start_sync(struct btrfs_root *root,
					    void __user *argp)
4275 4276 4277
{
	struct btrfs_trans_handle *trans;
	u64 transid;
T
Tsutomu Itoh 已提交
4278
	int ret;
4279

M
Miao Xie 已提交
4280
	trans = btrfs_attach_transaction_barrier(root);
4281 4282 4283 4284 4285 4286 4287 4288
	if (IS_ERR(trans)) {
		if (PTR_ERR(trans) != -ENOENT)
			return PTR_ERR(trans);

		/* No running transaction, don't bother */
		transid = root->fs_info->last_trans_committed;
		goto out;
	}
4289
	transid = trans->transid;
4290
	ret = btrfs_commit_transaction_async(trans, 0);
4291
	if (ret) {
4292
		btrfs_end_transaction(trans);
T
Tsutomu Itoh 已提交
4293
		return ret;
4294
	}
4295
out:
4296 4297 4298 4299 4300 4301
	if (argp)
		if (copy_to_user(argp, &transid, sizeof(transid)))
			return -EFAULT;
	return 0;
}

4302
static noinline long btrfs_ioctl_wait_sync(struct btrfs_fs_info *fs_info,
4303
					   void __user *argp)
4304 4305 4306 4307 4308 4309 4310 4311 4312
{
	u64 transid;

	if (argp) {
		if (copy_from_user(&transid, argp, sizeof(transid)))
			return -EFAULT;
	} else {
		transid = 0;  /* current trans */
	}
4313
	return btrfs_wait_for_commit(fs_info, transid);
4314 4315
}

M
Miao Xie 已提交
4316
static long btrfs_ioctl_scrub(struct file *file, void __user *arg)
J
Jan Schmidt 已提交
4317
{
4318
	struct btrfs_fs_info *fs_info = btrfs_sb(file_inode(file)->i_sb);
J
Jan Schmidt 已提交
4319
	struct btrfs_ioctl_scrub_args *sa;
M
Miao Xie 已提交
4320
	int ret;
J
Jan Schmidt 已提交
4321 4322 4323 4324 4325 4326 4327 4328

	if (!capable(CAP_SYS_ADMIN))
		return -EPERM;

	sa = memdup_user(arg, sizeof(*sa));
	if (IS_ERR(sa))
		return PTR_ERR(sa);

M
Miao Xie 已提交
4329 4330 4331 4332 4333 4334
	if (!(sa->flags & BTRFS_SCRUB_READONLY)) {
		ret = mnt_want_write_file(file);
		if (ret)
			goto out;
	}

4335
	ret = btrfs_scrub_dev(fs_info, sa->devid, sa->start, sa->end,
4336 4337
			      &sa->progress, sa->flags & BTRFS_SCRUB_READONLY,
			      0);
J
Jan Schmidt 已提交
4338 4339 4340 4341

	if (copy_to_user(arg, sa, sizeof(*sa)))
		ret = -EFAULT;

M
Miao Xie 已提交
4342 4343 4344
	if (!(sa->flags & BTRFS_SCRUB_READONLY))
		mnt_drop_write_file(file);
out:
J
Jan Schmidt 已提交
4345 4346 4347 4348
	kfree(sa);
	return ret;
}

4349
static long btrfs_ioctl_scrub_cancel(struct btrfs_fs_info *fs_info)
J
Jan Schmidt 已提交
4350 4351 4352 4353
{
	if (!capable(CAP_SYS_ADMIN))
		return -EPERM;

4354
	return btrfs_scrub_cancel(fs_info);
J
Jan Schmidt 已提交
4355 4356
}

4357
static long btrfs_ioctl_scrub_progress(struct btrfs_fs_info *fs_info,
J
Jan Schmidt 已提交
4358 4359 4360 4361 4362 4363 4364 4365 4366 4367 4368 4369
				       void __user *arg)
{
	struct btrfs_ioctl_scrub_args *sa;
	int ret;

	if (!capable(CAP_SYS_ADMIN))
		return -EPERM;

	sa = memdup_user(arg, sizeof(*sa));
	if (IS_ERR(sa))
		return PTR_ERR(sa);

4370
	ret = btrfs_scrub_progress(fs_info, sa->devid, &sa->progress);
J
Jan Schmidt 已提交
4371 4372 4373 4374 4375 4376 4377 4378

	if (copy_to_user(arg, sa, sizeof(*sa)))
		ret = -EFAULT;

	kfree(sa);
	return ret;
}

4379
static long btrfs_ioctl_get_dev_stats(struct btrfs_fs_info *fs_info,
4380
				      void __user *arg)
4381 4382 4383 4384 4385 4386 4387 4388
{
	struct btrfs_ioctl_get_dev_stats *sa;
	int ret;

	sa = memdup_user(arg, sizeof(*sa));
	if (IS_ERR(sa))
		return PTR_ERR(sa);

4389 4390 4391 4392 4393
	if ((sa->flags & BTRFS_DEV_STATS_RESET) && !capable(CAP_SYS_ADMIN)) {
		kfree(sa);
		return -EPERM;
	}

4394
	ret = btrfs_get_dev_stats(fs_info, sa);
4395 4396 4397 4398 4399 4400 4401 4402

	if (copy_to_user(arg, sa, sizeof(*sa)))
		ret = -EFAULT;

	kfree(sa);
	return ret;
}

4403 4404
static long btrfs_ioctl_dev_replace(struct btrfs_fs_info *fs_info,
				    void __user *arg)
4405 4406 4407 4408 4409 4410 4411 4412 4413 4414 4415 4416 4417
{
	struct btrfs_ioctl_dev_replace_args *p;
	int ret;

	if (!capable(CAP_SYS_ADMIN))
		return -EPERM;

	p = memdup_user(arg, sizeof(*p));
	if (IS_ERR(p))
		return PTR_ERR(p);

	switch (p->cmd) {
	case BTRFS_IOCTL_DEV_REPLACE_CMD_START:
4418
		if (sb_rdonly(fs_info->sb)) {
4419 4420 4421
			ret = -EROFS;
			goto out;
		}
4422
		if (test_and_set_bit(BTRFS_FS_EXCL_OP, &fs_info->flags)) {
4423
			ret = BTRFS_ERROR_DEV_EXCL_RUN_IN_PROGRESS;
4424
		} else {
4425
			ret = btrfs_dev_replace_by_ioctl(fs_info, p);
4426
			clear_bit(BTRFS_FS_EXCL_OP, &fs_info->flags);
4427 4428 4429
		}
		break;
	case BTRFS_IOCTL_DEV_REPLACE_CMD_STATUS:
4430
		btrfs_dev_replace_status(fs_info, p);
4431 4432 4433
		ret = 0;
		break;
	case BTRFS_IOCTL_DEV_REPLACE_CMD_CANCEL:
4434
		ret = btrfs_dev_replace_cancel(fs_info, p);
4435 4436 4437 4438 4439 4440 4441 4442
		break;
	default:
		ret = -EINVAL;
		break;
	}

	if (copy_to_user(arg, p, sizeof(*p)))
		ret = -EFAULT;
4443
out:
4444 4445 4446 4447
	kfree(p);
	return ret;
}

4448 4449 4450 4451
static long btrfs_ioctl_ino_to_path(struct btrfs_root *root, void __user *arg)
{
	int ret = 0;
	int i;
4452
	u64 rel_ptr;
4453
	int size;
4454
	struct btrfs_ioctl_ino_path_args *ipa = NULL;
4455 4456 4457
	struct inode_fs_paths *ipath = NULL;
	struct btrfs_path *path;

4458
	if (!capable(CAP_DAC_READ_SEARCH))
4459 4460 4461 4462 4463 4464 4465 4466 4467 4468 4469 4470 4471 4472 4473 4474 4475 4476 4477 4478 4479 4480 4481 4482 4483 4484 4485 4486
		return -EPERM;

	path = btrfs_alloc_path();
	if (!path) {
		ret = -ENOMEM;
		goto out;
	}

	ipa = memdup_user(arg, sizeof(*ipa));
	if (IS_ERR(ipa)) {
		ret = PTR_ERR(ipa);
		ipa = NULL;
		goto out;
	}

	size = min_t(u32, ipa->size, 4096);
	ipath = init_ipath(size, root, path);
	if (IS_ERR(ipath)) {
		ret = PTR_ERR(ipath);
		ipath = NULL;
		goto out;
	}

	ret = paths_from_inode(ipa->inum, ipath);
	if (ret < 0)
		goto out;

	for (i = 0; i < ipath->fspath->elem_cnt; ++i) {
4487 4488
		rel_ptr = ipath->fspath->val[i] -
			  (u64)(unsigned long)ipath->fspath->val;
4489
		ipath->fspath->val[i] = rel_ptr;
4490 4491
	}

4492 4493
	ret = copy_to_user((void *)(unsigned long)ipa->fspath,
			   (void *)(unsigned long)ipath->fspath, size);
4494 4495 4496 4497 4498 4499 4500 4501 4502 4503 4504 4505 4506 4507 4508 4509 4510 4511 4512 4513 4514 4515 4516 4517 4518 4519 4520 4521 4522 4523 4524 4525 4526
	if (ret) {
		ret = -EFAULT;
		goto out;
	}

out:
	btrfs_free_path(path);
	free_ipath(ipath);
	kfree(ipa);

	return ret;
}

static int build_ino_list(u64 inum, u64 offset, u64 root, void *ctx)
{
	struct btrfs_data_container *inodes = ctx;
	const size_t c = 3 * sizeof(u64);

	if (inodes->bytes_left >= c) {
		inodes->bytes_left -= c;
		inodes->val[inodes->elem_cnt] = inum;
		inodes->val[inodes->elem_cnt + 1] = offset;
		inodes->val[inodes->elem_cnt + 2] = root;
		inodes->elem_cnt += 3;
	} else {
		inodes->bytes_missing += c - inodes->bytes_left;
		inodes->bytes_left = 0;
		inodes->elem_missed += 3;
	}

	return 0;
}

4527
static long btrfs_ioctl_logical_to_ino(struct btrfs_fs_info *fs_info,
4528 4529 4530 4531 4532 4533 4534 4535 4536 4537 4538 4539
					void __user *arg)
{
	int ret = 0;
	int size;
	struct btrfs_ioctl_logical_ino_args *loi;
	struct btrfs_data_container *inodes = NULL;
	struct btrfs_path *path = NULL;

	if (!capable(CAP_SYS_ADMIN))
		return -EPERM;

	loi = memdup_user(arg, sizeof(*loi));
4540 4541
	if (IS_ERR(loi))
		return PTR_ERR(loi);
4542 4543 4544 4545 4546 4547 4548

	path = btrfs_alloc_path();
	if (!path) {
		ret = -ENOMEM;
		goto out;
	}

4549
	size = min_t(u32, loi->size, SZ_64K);
4550 4551 4552 4553 4554 4555 4556
	inodes = init_data_container(size);
	if (IS_ERR(inodes)) {
		ret = PTR_ERR(inodes);
		inodes = NULL;
		goto out;
	}

4557
	ret = iterate_inodes_from_logical(loi->logical, fs_info, path,
L
Liu Bo 已提交
4558 4559
					  build_ino_list, inodes);
	if (ret == -EINVAL)
4560 4561 4562 4563
		ret = -ENOENT;
	if (ret < 0)
		goto out;

4564 4565
	ret = copy_to_user((void *)(unsigned long)loi->inodes,
			   (void *)(unsigned long)inodes, size);
4566 4567 4568 4569 4570
	if (ret)
		ret = -EFAULT;

out:
	btrfs_free_path(path);
4571
	kvfree(inodes);
4572 4573 4574 4575 4576
	kfree(loi);

	return ret;
}

4577
void update_ioctl_balance_args(struct btrfs_fs_info *fs_info, int lock,
4578 4579 4580 4581 4582 4583
			       struct btrfs_ioctl_balance_args *bargs)
{
	struct btrfs_balance_control *bctl = fs_info->balance_ctl;

	bargs->flags = bctl->flags;

4584 4585 4586 4587
	if (atomic_read(&fs_info->balance_running))
		bargs->state |= BTRFS_BALANCE_STATE_RUNNING;
	if (atomic_read(&fs_info->balance_pause_req))
		bargs->state |= BTRFS_BALANCE_STATE_PAUSE_REQ;
4588 4589
	if (atomic_read(&fs_info->balance_cancel_req))
		bargs->state |= BTRFS_BALANCE_STATE_CANCEL_REQ;
4590

4591 4592 4593
	memcpy(&bargs->data, &bctl->data, sizeof(bargs->data));
	memcpy(&bargs->meta, &bctl->meta, sizeof(bargs->meta));
	memcpy(&bargs->sys, &bctl->sys, sizeof(bargs->sys));
4594 4595 4596 4597 4598 4599 4600 4601

	if (lock) {
		spin_lock(&fs_info->balance_lock);
		memcpy(&bargs->stat, &bctl->stat, sizeof(bargs->stat));
		spin_unlock(&fs_info->balance_lock);
	} else {
		memcpy(&bargs->stat, &bctl->stat, sizeof(bargs->stat));
	}
4602 4603
}

4604
static long btrfs_ioctl_balance(struct file *file, void __user *arg)
4605
{
A
Al Viro 已提交
4606
	struct btrfs_root *root = BTRFS_I(file_inode(file))->root;
4607 4608 4609
	struct btrfs_fs_info *fs_info = root->fs_info;
	struct btrfs_ioctl_balance_args *bargs;
	struct btrfs_balance_control *bctl;
4610
	bool need_unlock; /* for mut. excl. ops lock */
4611 4612 4613 4614 4615
	int ret;

	if (!capable(CAP_SYS_ADMIN))
		return -EPERM;

4616
	ret = mnt_want_write_file(file);
4617 4618 4619
	if (ret)
		return ret;

4620
again:
4621
	if (!test_and_set_bit(BTRFS_FS_EXCL_OP, &fs_info->flags)) {
4622 4623 4624 4625 4626 4627 4628
		mutex_lock(&fs_info->volume_mutex);
		mutex_lock(&fs_info->balance_mutex);
		need_unlock = true;
		goto locked;
	}

	/*
4629
	 * mut. excl. ops lock is locked.  Three possibilities:
4630 4631 4632 4633
	 *   (1) some other op is running
	 *   (2) balance is running
	 *   (3) balance is paused -- special case (think resume)
	 */
4634
	mutex_lock(&fs_info->balance_mutex);
4635 4636 4637 4638 4639 4640 4641 4642 4643 4644 4645 4646 4647 4648 4649 4650 4651 4652 4653 4654 4655 4656 4657 4658 4659 4660 4661
	if (fs_info->balance_ctl) {
		/* this is either (2) or (3) */
		if (!atomic_read(&fs_info->balance_running)) {
			mutex_unlock(&fs_info->balance_mutex);
			if (!mutex_trylock(&fs_info->volume_mutex))
				goto again;
			mutex_lock(&fs_info->balance_mutex);

			if (fs_info->balance_ctl &&
			    !atomic_read(&fs_info->balance_running)) {
				/* this is (3) */
				need_unlock = false;
				goto locked;
			}

			mutex_unlock(&fs_info->balance_mutex);
			mutex_unlock(&fs_info->volume_mutex);
			goto again;
		} else {
			/* this is (2) */
			mutex_unlock(&fs_info->balance_mutex);
			ret = -EINPROGRESS;
			goto out;
		}
	} else {
		/* this is (1) */
		mutex_unlock(&fs_info->balance_mutex);
4662
		ret = BTRFS_ERROR_DEV_EXCL_RUN_IN_PROGRESS;
4663 4664 4665 4666
		goto out;
	}

locked:
4667
	BUG_ON(!test_bit(BTRFS_FS_EXCL_OP, &fs_info->flags));
4668 4669 4670 4671 4672

	if (arg) {
		bargs = memdup_user(arg, sizeof(*bargs));
		if (IS_ERR(bargs)) {
			ret = PTR_ERR(bargs);
4673
			goto out_unlock;
4674
		}
4675 4676 4677 4678 4679 4680 4681 4682 4683 4684 4685 4686 4687 4688

		if (bargs->flags & BTRFS_BALANCE_RESUME) {
			if (!fs_info->balance_ctl) {
				ret = -ENOTCONN;
				goto out_bargs;
			}

			bctl = fs_info->balance_ctl;
			spin_lock(&fs_info->balance_lock);
			bctl->flags |= BTRFS_BALANCE_RESUME;
			spin_unlock(&fs_info->balance_lock);

			goto do_balance;
		}
4689 4690 4691 4692
	} else {
		bargs = NULL;
	}

4693
	if (fs_info->balance_ctl) {
4694 4695 4696 4697
		ret = -EINPROGRESS;
		goto out_bargs;
	}

4698
	bctl = kzalloc(sizeof(*bctl), GFP_KERNEL);
4699 4700 4701 4702 4703 4704 4705 4706 4707 4708 4709 4710
	if (!bctl) {
		ret = -ENOMEM;
		goto out_bargs;
	}

	bctl->fs_info = fs_info;
	if (arg) {
		memcpy(&bctl->data, &bargs->data, sizeof(bctl->data));
		memcpy(&bctl->meta, &bargs->meta, sizeof(bctl->meta));
		memcpy(&bctl->sys, &bargs->sys, sizeof(bctl->sys));

		bctl->flags = bargs->flags;
4711 4712 4713
	} else {
		/* balance everything - no filters */
		bctl->flags |= BTRFS_BALANCE_TYPE_MASK;
4714 4715
	}

4716 4717
	if (bctl->flags & ~(BTRFS_BALANCE_ARGS_MASK | BTRFS_BALANCE_TYPE_MASK)) {
		ret = -EINVAL;
4718
		goto out_bctl;
4719 4720
	}

4721
do_balance:
4722
	/*
4723
	 * Ownership of bctl and filesystem flag BTRFS_FS_EXCL_OP
4724 4725
	 * goes to to btrfs_balance.  bctl is freed in __cancel_balance,
	 * or, if restriper was paused all the way until unmount, in
4726
	 * free_fs_info.  The flag is cleared in __cancel_balance.
4727
	 */
4728 4729 4730
	need_unlock = false;

	ret = btrfs_balance(bctl, bargs);
4731
	bctl = NULL;
4732

4733 4734 4735 4736 4737
	if (arg) {
		if (copy_to_user(arg, bargs, sizeof(*bargs)))
			ret = -EFAULT;
	}

4738 4739
out_bctl:
	kfree(bctl);
4740 4741
out_bargs:
	kfree(bargs);
4742
out_unlock:
4743 4744
	mutex_unlock(&fs_info->balance_mutex);
	mutex_unlock(&fs_info->volume_mutex);
4745
	if (need_unlock)
4746
		clear_bit(BTRFS_FS_EXCL_OP, &fs_info->flags);
4747
out:
4748
	mnt_drop_write_file(file);
4749 4750 4751
	return ret;
}

4752
static long btrfs_ioctl_balance_ctl(struct btrfs_fs_info *fs_info, int cmd)
4753 4754 4755 4756 4757 4758
{
	if (!capable(CAP_SYS_ADMIN))
		return -EPERM;

	switch (cmd) {
	case BTRFS_BALANCE_CTL_PAUSE:
4759
		return btrfs_pause_balance(fs_info);
4760
	case BTRFS_BALANCE_CTL_CANCEL:
4761
		return btrfs_cancel_balance(fs_info);
4762 4763 4764 4765 4766
	}

	return -EINVAL;
}

4767
static long btrfs_ioctl_balance_progress(struct btrfs_fs_info *fs_info,
4768 4769 4770 4771 4772 4773 4774 4775 4776 4777 4778 4779 4780 4781
					 void __user *arg)
{
	struct btrfs_ioctl_balance_args *bargs;
	int ret = 0;

	if (!capable(CAP_SYS_ADMIN))
		return -EPERM;

	mutex_lock(&fs_info->balance_mutex);
	if (!fs_info->balance_ctl) {
		ret = -ENOTCONN;
		goto out;
	}

4782
	bargs = kzalloc(sizeof(*bargs), GFP_KERNEL);
4783 4784 4785 4786 4787 4788 4789 4790 4791 4792 4793 4794 4795 4796 4797 4798
	if (!bargs) {
		ret = -ENOMEM;
		goto out;
	}

	update_ioctl_balance_args(fs_info, 1, bargs);

	if (copy_to_user(arg, bargs, sizeof(*bargs)))
		ret = -EFAULT;

	kfree(bargs);
out:
	mutex_unlock(&fs_info->balance_mutex);
	return ret;
}

4799
static long btrfs_ioctl_quota_ctl(struct file *file, void __user *arg)
A
Arne Jansen 已提交
4800
{
4801 4802
	struct inode *inode = file_inode(file);
	struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb);
A
Arne Jansen 已提交
4803 4804 4805 4806 4807 4808 4809 4810
	struct btrfs_ioctl_quota_ctl_args *sa;
	struct btrfs_trans_handle *trans = NULL;
	int ret;
	int err;

	if (!capable(CAP_SYS_ADMIN))
		return -EPERM;

4811 4812 4813
	ret = mnt_want_write_file(file);
	if (ret)
		return ret;
A
Arne Jansen 已提交
4814 4815

	sa = memdup_user(arg, sizeof(*sa));
4816 4817 4818 4819
	if (IS_ERR(sa)) {
		ret = PTR_ERR(sa);
		goto drop_write;
	}
A
Arne Jansen 已提交
4820

4821 4822
	down_write(&fs_info->subvol_sem);
	trans = btrfs_start_transaction(fs_info->tree_root, 2);
J
Jan Schmidt 已提交
4823 4824 4825
	if (IS_ERR(trans)) {
		ret = PTR_ERR(trans);
		goto out;
A
Arne Jansen 已提交
4826 4827 4828 4829
	}

	switch (sa->cmd) {
	case BTRFS_QUOTA_CTL_ENABLE:
4830
		ret = btrfs_quota_enable(trans, fs_info);
A
Arne Jansen 已提交
4831 4832
		break;
	case BTRFS_QUOTA_CTL_DISABLE:
4833
		ret = btrfs_quota_disable(trans, fs_info);
A
Arne Jansen 已提交
4834 4835 4836 4837 4838 4839
		break;
	default:
		ret = -EINVAL;
		break;
	}

4840
	err = btrfs_commit_transaction(trans);
J
Jan Schmidt 已提交
4841 4842
	if (err && !ret)
		ret = err;
A
Arne Jansen 已提交
4843 4844
out:
	kfree(sa);
4845
	up_write(&fs_info->subvol_sem);
4846 4847
drop_write:
	mnt_drop_write_file(file);
A
Arne Jansen 已提交
4848 4849 4850
	return ret;
}

4851
static long btrfs_ioctl_qgroup_assign(struct file *file, void __user *arg)
A
Arne Jansen 已提交
4852
{
4853 4854 4855
	struct inode *inode = file_inode(file);
	struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb);
	struct btrfs_root *root = BTRFS_I(inode)->root;
A
Arne Jansen 已提交
4856 4857 4858 4859 4860 4861 4862 4863
	struct btrfs_ioctl_qgroup_assign_args *sa;
	struct btrfs_trans_handle *trans;
	int ret;
	int err;

	if (!capable(CAP_SYS_ADMIN))
		return -EPERM;

4864 4865 4866
	ret = mnt_want_write_file(file);
	if (ret)
		return ret;
A
Arne Jansen 已提交
4867 4868

	sa = memdup_user(arg, sizeof(*sa));
4869 4870 4871 4872
	if (IS_ERR(sa)) {
		ret = PTR_ERR(sa);
		goto drop_write;
	}
A
Arne Jansen 已提交
4873 4874 4875 4876 4877 4878 4879 4880

	trans = btrfs_join_transaction(root);
	if (IS_ERR(trans)) {
		ret = PTR_ERR(trans);
		goto out;
	}

	if (sa->assign) {
4881
		ret = btrfs_add_qgroup_relation(trans, fs_info,
A
Arne Jansen 已提交
4882 4883
						sa->src, sa->dst);
	} else {
4884
		ret = btrfs_del_qgroup_relation(trans, fs_info,
A
Arne Jansen 已提交
4885 4886 4887
						sa->src, sa->dst);
	}

4888
	/* update qgroup status and info */
4889
	err = btrfs_run_qgroups(trans, fs_info);
4890
	if (err < 0)
4891 4892
		btrfs_handle_fs_error(fs_info, err,
				      "failed to update qgroup status and info");
4893
	err = btrfs_end_transaction(trans);
A
Arne Jansen 已提交
4894 4895 4896 4897 4898
	if (err && !ret)
		ret = err;

out:
	kfree(sa);
4899 4900
drop_write:
	mnt_drop_write_file(file);
A
Arne Jansen 已提交
4901 4902 4903
	return ret;
}

4904
static long btrfs_ioctl_qgroup_create(struct file *file, void __user *arg)
A
Arne Jansen 已提交
4905
{
4906 4907 4908
	struct inode *inode = file_inode(file);
	struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb);
	struct btrfs_root *root = BTRFS_I(inode)->root;
A
Arne Jansen 已提交
4909 4910 4911 4912 4913 4914 4915 4916
	struct btrfs_ioctl_qgroup_create_args *sa;
	struct btrfs_trans_handle *trans;
	int ret;
	int err;

	if (!capable(CAP_SYS_ADMIN))
		return -EPERM;

4917 4918 4919
	ret = mnt_want_write_file(file);
	if (ret)
		return ret;
A
Arne Jansen 已提交
4920 4921

	sa = memdup_user(arg, sizeof(*sa));
4922 4923 4924 4925
	if (IS_ERR(sa)) {
		ret = PTR_ERR(sa);
		goto drop_write;
	}
A
Arne Jansen 已提交
4926

M
Miao Xie 已提交
4927 4928 4929 4930 4931
	if (!sa->qgroupid) {
		ret = -EINVAL;
		goto out;
	}

A
Arne Jansen 已提交
4932 4933 4934 4935 4936 4937 4938
	trans = btrfs_join_transaction(root);
	if (IS_ERR(trans)) {
		ret = PTR_ERR(trans);
		goto out;
	}

	if (sa->create) {
4939
		ret = btrfs_create_qgroup(trans, fs_info, sa->qgroupid);
A
Arne Jansen 已提交
4940
	} else {
4941
		ret = btrfs_remove_qgroup(trans, fs_info, sa->qgroupid);
A
Arne Jansen 已提交
4942 4943
	}

4944
	err = btrfs_end_transaction(trans);
A
Arne Jansen 已提交
4945 4946 4947 4948 4949
	if (err && !ret)
		ret = err;

out:
	kfree(sa);
4950 4951
drop_write:
	mnt_drop_write_file(file);
A
Arne Jansen 已提交
4952 4953 4954
	return ret;
}

4955
static long btrfs_ioctl_qgroup_limit(struct file *file, void __user *arg)
A
Arne Jansen 已提交
4956
{
4957 4958 4959
	struct inode *inode = file_inode(file);
	struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb);
	struct btrfs_root *root = BTRFS_I(inode)->root;
A
Arne Jansen 已提交
4960 4961 4962 4963 4964 4965 4966 4967 4968
	struct btrfs_ioctl_qgroup_limit_args *sa;
	struct btrfs_trans_handle *trans;
	int ret;
	int err;
	u64 qgroupid;

	if (!capable(CAP_SYS_ADMIN))
		return -EPERM;

4969 4970 4971
	ret = mnt_want_write_file(file);
	if (ret)
		return ret;
A
Arne Jansen 已提交
4972 4973

	sa = memdup_user(arg, sizeof(*sa));
4974 4975 4976 4977
	if (IS_ERR(sa)) {
		ret = PTR_ERR(sa);
		goto drop_write;
	}
A
Arne Jansen 已提交
4978 4979 4980 4981 4982 4983 4984 4985 4986 4987 4988 4989 4990

	trans = btrfs_join_transaction(root);
	if (IS_ERR(trans)) {
		ret = PTR_ERR(trans);
		goto out;
	}

	qgroupid = sa->qgroupid;
	if (!qgroupid) {
		/* take the current subvol as qgroup */
		qgroupid = root->root_key.objectid;
	}

4991
	ret = btrfs_limit_qgroup(trans, fs_info, qgroupid, &sa->lim);
A
Arne Jansen 已提交
4992

4993
	err = btrfs_end_transaction(trans);
A
Arne Jansen 已提交
4994 4995 4996 4997 4998
	if (err && !ret)
		ret = err;

out:
	kfree(sa);
4999 5000
drop_write:
	mnt_drop_write_file(file);
A
Arne Jansen 已提交
5001 5002 5003
	return ret;
}

J
Jan Schmidt 已提交
5004 5005
static long btrfs_ioctl_quota_rescan(struct file *file, void __user *arg)
{
5006 5007
	struct inode *inode = file_inode(file);
	struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb);
J
Jan Schmidt 已提交
5008 5009 5010 5011 5012 5013 5014 5015 5016 5017 5018 5019 5020 5021 5022 5023 5024 5025 5026 5027 5028
	struct btrfs_ioctl_quota_rescan_args *qsa;
	int ret;

	if (!capable(CAP_SYS_ADMIN))
		return -EPERM;

	ret = mnt_want_write_file(file);
	if (ret)
		return ret;

	qsa = memdup_user(arg, sizeof(*qsa));
	if (IS_ERR(qsa)) {
		ret = PTR_ERR(qsa);
		goto drop_write;
	}

	if (qsa->flags) {
		ret = -EINVAL;
		goto out;
	}

5029
	ret = btrfs_qgroup_rescan(fs_info);
J
Jan Schmidt 已提交
5030 5031 5032 5033 5034 5035 5036 5037 5038 5039

out:
	kfree(qsa);
drop_write:
	mnt_drop_write_file(file);
	return ret;
}

static long btrfs_ioctl_quota_rescan_status(struct file *file, void __user *arg)
{
5040 5041
	struct inode *inode = file_inode(file);
	struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb);
J
Jan Schmidt 已提交
5042 5043 5044 5045 5046 5047
	struct btrfs_ioctl_quota_rescan_args *qsa;
	int ret = 0;

	if (!capable(CAP_SYS_ADMIN))
		return -EPERM;

5048
	qsa = kzalloc(sizeof(*qsa), GFP_KERNEL);
J
Jan Schmidt 已提交
5049 5050 5051
	if (!qsa)
		return -ENOMEM;

5052
	if (fs_info->qgroup_flags & BTRFS_QGROUP_STATUS_FLAG_RESCAN) {
J
Jan Schmidt 已提交
5053
		qsa->flags = 1;
5054
		qsa->progress = fs_info->qgroup_rescan_progress.objectid;
J
Jan Schmidt 已提交
5055 5056 5057 5058 5059 5060 5061 5062 5063
	}

	if (copy_to_user(arg, qsa, sizeof(*qsa)))
		ret = -EFAULT;

	kfree(qsa);
	return ret;
}

5064 5065
static long btrfs_ioctl_quota_rescan_wait(struct file *file, void __user *arg)
{
5066 5067
	struct inode *inode = file_inode(file);
	struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb);
5068 5069 5070 5071

	if (!capable(CAP_SYS_ADMIN))
		return -EPERM;

5072
	return btrfs_qgroup_wait_for_completion(fs_info, true);
5073 5074
}

5075 5076
static long _btrfs_ioctl_set_received_subvol(struct file *file,
					    struct btrfs_ioctl_received_subvol_args *sa)
5077
{
A
Al Viro 已提交
5078
	struct inode *inode = file_inode(file);
5079
	struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb);
5080 5081 5082
	struct btrfs_root *root = BTRFS_I(inode)->root;
	struct btrfs_root_item *root_item = &root->root_item;
	struct btrfs_trans_handle *trans;
5083
	struct timespec ct = current_time(inode);
5084
	int ret = 0;
5085
	int received_uuid_changed;
5086

5087 5088 5089
	if (!inode_owner_or_capable(inode))
		return -EPERM;

5090 5091 5092 5093
	ret = mnt_want_write_file(file);
	if (ret < 0)
		return ret;

5094
	down_write(&fs_info->subvol_sem);
5095

5096
	if (btrfs_ino(BTRFS_I(inode)) != BTRFS_FIRST_FREE_OBJECTID) {
5097 5098 5099 5100 5101 5102 5103 5104 5105
		ret = -EINVAL;
		goto out;
	}

	if (btrfs_root_readonly(root)) {
		ret = -EROFS;
		goto out;
	}

5106 5107 5108 5109 5110
	/*
	 * 1 - root item
	 * 2 - uuid items (received uuid + subvol uuid)
	 */
	trans = btrfs_start_transaction(root, 3);
5111 5112 5113 5114 5115 5116 5117 5118 5119 5120
	if (IS_ERR(trans)) {
		ret = PTR_ERR(trans);
		trans = NULL;
		goto out;
	}

	sa->rtransid = trans->transid;
	sa->rtime.sec = ct.tv_sec;
	sa->rtime.nsec = ct.tv_nsec;

5121 5122 5123 5124
	received_uuid_changed = memcmp(root_item->received_uuid, sa->uuid,
				       BTRFS_UUID_SIZE);
	if (received_uuid_changed &&
	    !btrfs_is_empty_uuid(root_item->received_uuid))
5125
		btrfs_uuid_tree_rem(trans, fs_info, root_item->received_uuid,
5126 5127
				    BTRFS_UUID_KEY_RECEIVED_SUBVOL,
				    root->root_key.objectid);
5128 5129 5130
	memcpy(root_item->received_uuid, sa->uuid, BTRFS_UUID_SIZE);
	btrfs_set_root_stransid(root_item, sa->stransid);
	btrfs_set_root_rtransid(root_item, sa->rtransid);
5131 5132 5133 5134
	btrfs_set_stack_timespec_sec(&root_item->stime, sa->stime.sec);
	btrfs_set_stack_timespec_nsec(&root_item->stime, sa->stime.nsec);
	btrfs_set_stack_timespec_sec(&root_item->rtime, sa->rtime.sec);
	btrfs_set_stack_timespec_nsec(&root_item->rtime, sa->rtime.nsec);
5135

5136
	ret = btrfs_update_root(trans, fs_info->tree_root,
5137 5138
				&root->root_key, &root->root_item);
	if (ret < 0) {
5139
		btrfs_end_transaction(trans);
5140
		goto out;
5141 5142
	}
	if (received_uuid_changed && !btrfs_is_empty_uuid(sa->uuid)) {
5143
		ret = btrfs_uuid_tree_add(trans, fs_info, sa->uuid,
5144 5145 5146
					  BTRFS_UUID_KEY_RECEIVED_SUBVOL,
					  root->root_key.objectid);
		if (ret < 0 && ret != -EEXIST) {
5147
			btrfs_abort_transaction(trans, ret);
5148
			goto out;
5149 5150
		}
	}
5151
	ret = btrfs_commit_transaction(trans);
5152
	if (ret < 0) {
5153
		btrfs_abort_transaction(trans, ret);
5154
		goto out;
5155 5156
	}

5157
out:
5158
	up_write(&fs_info->subvol_sem);
5159 5160 5161 5162 5163 5164 5165 5166 5167 5168 5169 5170 5171
	mnt_drop_write_file(file);
	return ret;
}

#ifdef CONFIG_64BIT
static long btrfs_ioctl_set_received_subvol_32(struct file *file,
						void __user *arg)
{
	struct btrfs_ioctl_received_subvol_args_32 *args32 = NULL;
	struct btrfs_ioctl_received_subvol_args *args64 = NULL;
	int ret = 0;

	args32 = memdup_user(arg, sizeof(*args32));
5172 5173
	if (IS_ERR(args32))
		return PTR_ERR(args32);
5174

5175
	args64 = kmalloc(sizeof(*args64), GFP_KERNEL);
5176 5177
	if (!args64) {
		ret = -ENOMEM;
5178 5179 5180 5181 5182 5183 5184 5185 5186 5187 5188 5189 5190 5191 5192 5193 5194 5195 5196 5197 5198 5199 5200 5201 5202 5203 5204 5205 5206 5207 5208 5209 5210 5211 5212 5213 5214 5215 5216 5217 5218 5219 5220
		goto out;
	}

	memcpy(args64->uuid, args32->uuid, BTRFS_UUID_SIZE);
	args64->stransid = args32->stransid;
	args64->rtransid = args32->rtransid;
	args64->stime.sec = args32->stime.sec;
	args64->stime.nsec = args32->stime.nsec;
	args64->rtime.sec = args32->rtime.sec;
	args64->rtime.nsec = args32->rtime.nsec;
	args64->flags = args32->flags;

	ret = _btrfs_ioctl_set_received_subvol(file, args64);
	if (ret)
		goto out;

	memcpy(args32->uuid, args64->uuid, BTRFS_UUID_SIZE);
	args32->stransid = args64->stransid;
	args32->rtransid = args64->rtransid;
	args32->stime.sec = args64->stime.sec;
	args32->stime.nsec = args64->stime.nsec;
	args32->rtime.sec = args64->rtime.sec;
	args32->rtime.nsec = args64->rtime.nsec;
	args32->flags = args64->flags;

	ret = copy_to_user(arg, args32, sizeof(*args32));
	if (ret)
		ret = -EFAULT;

out:
	kfree(args32);
	kfree(args64);
	return ret;
}
#endif

static long btrfs_ioctl_set_received_subvol(struct file *file,
					    void __user *arg)
{
	struct btrfs_ioctl_received_subvol_args *sa = NULL;
	int ret = 0;

	sa = memdup_user(arg, sizeof(*sa));
5221 5222
	if (IS_ERR(sa))
		return PTR_ERR(sa);
5223 5224 5225 5226 5227 5228

	ret = _btrfs_ioctl_set_received_subvol(file, sa);

	if (ret)
		goto out;

5229 5230 5231 5232 5233 5234 5235 5236 5237
	ret = copy_to_user(arg, sa, sizeof(*sa));
	if (ret)
		ret = -EFAULT;

out:
	kfree(sa);
	return ret;
}

5238 5239
static int btrfs_ioctl_get_fslabel(struct file *file, void __user *arg)
{
5240 5241
	struct inode *inode = file_inode(file);
	struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb);
5242
	size_t len;
5243
	int ret;
5244 5245
	char label[BTRFS_LABEL_SIZE];

5246 5247 5248
	spin_lock(&fs_info->super_lock);
	memcpy(label, fs_info->super_copy->label, BTRFS_LABEL_SIZE);
	spin_unlock(&fs_info->super_lock);
5249 5250

	len = strnlen(label, BTRFS_LABEL_SIZE);
5251 5252

	if (len == BTRFS_LABEL_SIZE) {
5253 5254 5255
		btrfs_warn(fs_info,
			   "label is too long, return the first %zu bytes",
			   --len);
5256 5257 5258 5259 5260 5261 5262
	}

	ret = copy_to_user(arg, label, len);

	return ret ? -EFAULT : 0;
}

5263 5264
static int btrfs_ioctl_set_fslabel(struct file *file, void __user *arg)
{
5265 5266 5267 5268
	struct inode *inode = file_inode(file);
	struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb);
	struct btrfs_root *root = BTRFS_I(inode)->root;
	struct btrfs_super_block *super_block = fs_info->super_copy;
5269 5270 5271 5272 5273 5274 5275 5276 5277 5278 5279
	struct btrfs_trans_handle *trans;
	char label[BTRFS_LABEL_SIZE];
	int ret;

	if (!capable(CAP_SYS_ADMIN))
		return -EPERM;

	if (copy_from_user(label, arg, sizeof(label)))
		return -EFAULT;

	if (strnlen(label, BTRFS_LABEL_SIZE) == BTRFS_LABEL_SIZE) {
5280
		btrfs_err(fs_info,
J
Jeff Mahoney 已提交
5281 5282
			  "unable to set label with more than %d bytes",
			  BTRFS_LABEL_SIZE - 1);
5283 5284 5285 5286 5287 5288 5289 5290 5291 5292 5293 5294 5295
		return -EINVAL;
	}

	ret = mnt_want_write_file(file);
	if (ret)
		return ret;

	trans = btrfs_start_transaction(root, 0);
	if (IS_ERR(trans)) {
		ret = PTR_ERR(trans);
		goto out_unlock;
	}

5296
	spin_lock(&fs_info->super_lock);
5297
	strcpy(super_block->label, label);
5298
	spin_unlock(&fs_info->super_lock);
5299
	ret = btrfs_commit_transaction(trans);
5300 5301 5302 5303 5304 5305

out_unlock:
	mnt_drop_write_file(file);
	return ret;
}

5306 5307 5308 5309 5310
#define INIT_FEATURE_FLAGS(suffix) \
	{ .compat_flags = BTRFS_FEATURE_COMPAT_##suffix, \
	  .compat_ro_flags = BTRFS_FEATURE_COMPAT_RO_##suffix, \
	  .incompat_flags = BTRFS_FEATURE_INCOMPAT_##suffix }

5311
int btrfs_ioctl_get_supported_features(void __user *arg)
5312
{
D
David Sterba 已提交
5313
	static const struct btrfs_ioctl_feature_flags features[3] = {
5314 5315 5316 5317 5318 5319 5320 5321 5322 5323 5324 5325 5326
		INIT_FEATURE_FLAGS(SUPP),
		INIT_FEATURE_FLAGS(SAFE_SET),
		INIT_FEATURE_FLAGS(SAFE_CLEAR)
	};

	if (copy_to_user(arg, &features, sizeof(features)))
		return -EFAULT;

	return 0;
}

static int btrfs_ioctl_get_features(struct file *file, void __user *arg)
{
5327 5328 5329
	struct inode *inode = file_inode(file);
	struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb);
	struct btrfs_super_block *super_block = fs_info->super_copy;
5330 5331 5332 5333 5334 5335 5336 5337 5338 5339 5340 5341
	struct btrfs_ioctl_feature_flags features;

	features.compat_flags = btrfs_super_compat_flags(super_block);
	features.compat_ro_flags = btrfs_super_compat_ro_flags(super_block);
	features.incompat_flags = btrfs_super_incompat_flags(super_block);

	if (copy_to_user(arg, &features, sizeof(features)))
		return -EFAULT;

	return 0;
}

5342
static int check_feature_bits(struct btrfs_fs_info *fs_info,
5343
			      enum btrfs_feature_set set,
5344 5345 5346
			      u64 change_mask, u64 flags, u64 supported_flags,
			      u64 safe_set, u64 safe_clear)
{
5347 5348
	const char *type = btrfs_feature_set_names[set];
	char *names;
5349 5350 5351 5352 5353 5354
	u64 disallowed, unsupported;
	u64 set_mask = flags & change_mask;
	u64 clear_mask = ~flags & change_mask;

	unsupported = set_mask & ~supported_flags;
	if (unsupported) {
5355 5356
		names = btrfs_printable_features(set, unsupported);
		if (names) {
5357 5358 5359
			btrfs_warn(fs_info,
				   "this kernel does not support the %s feature bit%s",
				   names, strchr(names, ',') ? "s" : "");
5360 5361
			kfree(names);
		} else
5362 5363 5364
			btrfs_warn(fs_info,
				   "this kernel does not support %s bits 0x%llx",
				   type, unsupported);
5365 5366 5367 5368 5369
		return -EOPNOTSUPP;
	}

	disallowed = set_mask & ~safe_set;
	if (disallowed) {
5370 5371
		names = btrfs_printable_features(set, disallowed);
		if (names) {
5372 5373 5374
			btrfs_warn(fs_info,
				   "can't set the %s feature bit%s while mounted",
				   names, strchr(names, ',') ? "s" : "");
5375 5376
			kfree(names);
		} else
5377 5378 5379
			btrfs_warn(fs_info,
				   "can't set %s bits 0x%llx while mounted",
				   type, disallowed);
5380 5381 5382 5383 5384
		return -EPERM;
	}

	disallowed = clear_mask & ~safe_clear;
	if (disallowed) {
5385 5386
		names = btrfs_printable_features(set, disallowed);
		if (names) {
5387 5388 5389
			btrfs_warn(fs_info,
				   "can't clear the %s feature bit%s while mounted",
				   names, strchr(names, ',') ? "s" : "");
5390 5391
			kfree(names);
		} else
5392 5393 5394
			btrfs_warn(fs_info,
				   "can't clear %s bits 0x%llx while mounted",
				   type, disallowed);
5395 5396 5397 5398 5399 5400
		return -EPERM;
	}

	return 0;
}

5401 5402
#define check_feature(fs_info, change_mask, flags, mask_base)	\
check_feature_bits(fs_info, FEAT_##mask_base, change_mask, flags,	\
5403 5404 5405 5406 5407 5408
		   BTRFS_FEATURE_ ## mask_base ## _SUPP,	\
		   BTRFS_FEATURE_ ## mask_base ## _SAFE_SET,	\
		   BTRFS_FEATURE_ ## mask_base ## _SAFE_CLEAR)

static int btrfs_ioctl_set_features(struct file *file, void __user *arg)
{
5409 5410 5411 5412
	struct inode *inode = file_inode(file);
	struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb);
	struct btrfs_root *root = BTRFS_I(inode)->root;
	struct btrfs_super_block *super_block = fs_info->super_copy;
5413 5414 5415 5416 5417 5418 5419 5420 5421 5422 5423 5424 5425 5426 5427 5428
	struct btrfs_ioctl_feature_flags flags[2];
	struct btrfs_trans_handle *trans;
	u64 newflags;
	int ret;

	if (!capable(CAP_SYS_ADMIN))
		return -EPERM;

	if (copy_from_user(flags, arg, sizeof(flags)))
		return -EFAULT;

	/* Nothing to do */
	if (!flags[0].compat_flags && !flags[0].compat_ro_flags &&
	    !flags[0].incompat_flags)
		return 0;

5429
	ret = check_feature(fs_info, flags[0].compat_flags,
5430 5431 5432 5433
			    flags[1].compat_flags, COMPAT);
	if (ret)
		return ret;

5434
	ret = check_feature(fs_info, flags[0].compat_ro_flags,
5435 5436 5437 5438
			    flags[1].compat_ro_flags, COMPAT_RO);
	if (ret)
		return ret;

5439
	ret = check_feature(fs_info, flags[0].incompat_flags,
5440 5441 5442 5443
			    flags[1].incompat_flags, INCOMPAT);
	if (ret)
		return ret;

5444 5445 5446 5447
	ret = mnt_want_write_file(file);
	if (ret)
		return ret;

5448
	trans = btrfs_start_transaction(root, 0);
5449 5450 5451 5452
	if (IS_ERR(trans)) {
		ret = PTR_ERR(trans);
		goto out_drop_write;
	}
5453

5454
	spin_lock(&fs_info->super_lock);
5455 5456 5457 5458 5459 5460 5461 5462 5463 5464 5465 5466 5467 5468
	newflags = btrfs_super_compat_flags(super_block);
	newflags |= flags[0].compat_flags & flags[1].compat_flags;
	newflags &= ~(flags[0].compat_flags & ~flags[1].compat_flags);
	btrfs_set_super_compat_flags(super_block, newflags);

	newflags = btrfs_super_compat_ro_flags(super_block);
	newflags |= flags[0].compat_ro_flags & flags[1].compat_ro_flags;
	newflags &= ~(flags[0].compat_ro_flags & ~flags[1].compat_ro_flags);
	btrfs_set_super_compat_ro_flags(super_block, newflags);

	newflags = btrfs_super_incompat_flags(super_block);
	newflags |= flags[0].incompat_flags & flags[1].incompat_flags;
	newflags &= ~(flags[0].incompat_flags & ~flags[1].incompat_flags);
	btrfs_set_super_incompat_flags(super_block, newflags);
5469
	spin_unlock(&fs_info->super_lock);
5470

5471
	ret = btrfs_commit_transaction(trans);
5472 5473 5474 5475
out_drop_write:
	mnt_drop_write_file(file);

	return ret;
5476 5477
}

C
Christoph Hellwig 已提交
5478 5479 5480
long btrfs_ioctl(struct file *file, unsigned int
		cmd, unsigned long arg)
{
5481 5482 5483
	struct inode *inode = file_inode(file);
	struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb);
	struct btrfs_root *root = BTRFS_I(inode)->root;
5484
	void __user *argp = (void __user *)arg;
C
Christoph Hellwig 已提交
5485 5486

	switch (cmd) {
5487 5488 5489 5490 5491 5492
	case FS_IOC_GETFLAGS:
		return btrfs_ioctl_getflags(file, argp);
	case FS_IOC_SETFLAGS:
		return btrfs_ioctl_setflags(file, argp);
	case FS_IOC_GETVERSION:
		return btrfs_ioctl_getversion(file, argp);
5493 5494
	case FITRIM:
		return btrfs_ioctl_fitrim(file, argp);
C
Christoph Hellwig 已提交
5495
	case BTRFS_IOC_SNAP_CREATE:
5496
		return btrfs_ioctl_snap_create(file, argp, 0);
5497
	case BTRFS_IOC_SNAP_CREATE_V2:
5498
		return btrfs_ioctl_snap_create_v2(file, argp, 0);
5499
	case BTRFS_IOC_SUBVOL_CREATE:
5500
		return btrfs_ioctl_snap_create(file, argp, 1);
A
Arne Jansen 已提交
5501 5502
	case BTRFS_IOC_SUBVOL_CREATE_V2:
		return btrfs_ioctl_snap_create_v2(file, argp, 1);
5503 5504
	case BTRFS_IOC_SNAP_DESTROY:
		return btrfs_ioctl_snap_destroy(file, argp);
5505 5506 5507 5508
	case BTRFS_IOC_SUBVOL_GETFLAGS:
		return btrfs_ioctl_subvol_getflags(file, argp);
	case BTRFS_IOC_SUBVOL_SETFLAGS:
		return btrfs_ioctl_subvol_setflags(file, argp);
5509 5510
	case BTRFS_IOC_DEFAULT_SUBVOL:
		return btrfs_ioctl_default_subvol(file, argp);
C
Christoph Hellwig 已提交
5511
	case BTRFS_IOC_DEFRAG:
C
Chris Mason 已提交
5512 5513 5514
		return btrfs_ioctl_defrag(file, NULL);
	case BTRFS_IOC_DEFRAG_RANGE:
		return btrfs_ioctl_defrag(file, argp);
C
Christoph Hellwig 已提交
5515
	case BTRFS_IOC_RESIZE:
5516
		return btrfs_ioctl_resize(file, argp);
C
Christoph Hellwig 已提交
5517
	case BTRFS_IOC_ADD_DEV:
5518
		return btrfs_ioctl_add_dev(fs_info, argp);
C
Christoph Hellwig 已提交
5519
	case BTRFS_IOC_RM_DEV:
5520
		return btrfs_ioctl_rm_dev(file, argp);
5521 5522
	case BTRFS_IOC_RM_DEV_V2:
		return btrfs_ioctl_rm_dev_v2(file, argp);
J
Jan Schmidt 已提交
5523
	case BTRFS_IOC_FS_INFO:
5524
		return btrfs_ioctl_fs_info(fs_info, argp);
J
Jan Schmidt 已提交
5525
	case BTRFS_IOC_DEV_INFO:
5526
		return btrfs_ioctl_dev_info(fs_info, argp);
C
Christoph Hellwig 已提交
5527
	case BTRFS_IOC_BALANCE:
5528
		return btrfs_ioctl_balance(file, NULL);
C
Christoph Hellwig 已提交
5529 5530 5531 5532
	case BTRFS_IOC_TRANS_START:
		return btrfs_ioctl_trans_start(file);
	case BTRFS_IOC_TRANS_END:
		return btrfs_ioctl_trans_end(file);
5533 5534
	case BTRFS_IOC_TREE_SEARCH:
		return btrfs_ioctl_tree_search(file, argp);
G
Gerhard Heift 已提交
5535 5536
	case BTRFS_IOC_TREE_SEARCH_V2:
		return btrfs_ioctl_tree_search_v2(file, argp);
5537 5538
	case BTRFS_IOC_INO_LOOKUP:
		return btrfs_ioctl_ino_lookup(file, argp);
5539 5540 5541
	case BTRFS_IOC_INO_PATHS:
		return btrfs_ioctl_ino_to_path(root, argp);
	case BTRFS_IOC_LOGICAL_INO:
5542
		return btrfs_ioctl_logical_to_ino(fs_info, argp);
J
Josef Bacik 已提交
5543
	case BTRFS_IOC_SPACE_INFO:
5544
		return btrfs_ioctl_space_info(fs_info, argp);
5545 5546 5547
	case BTRFS_IOC_SYNC: {
		int ret;

5548
		ret = btrfs_start_delalloc_roots(fs_info, 0, -1);
5549 5550
		if (ret)
			return ret;
5551
		ret = btrfs_sync_fs(inode->i_sb, 1);
5552 5553
		/*
		 * The transaction thread may want to do more work,
5554
		 * namely it pokes the cleaner kthread that will start
5555 5556
		 * processing uncleaned subvols.
		 */
5557
		wake_up_process(fs_info->transaction_kthread);
5558 5559
		return ret;
	}
5560
	case BTRFS_IOC_START_SYNC:
5561
		return btrfs_ioctl_start_sync(root, argp);
5562
	case BTRFS_IOC_WAIT_SYNC:
5563
		return btrfs_ioctl_wait_sync(fs_info, argp);
J
Jan Schmidt 已提交
5564
	case BTRFS_IOC_SCRUB:
M
Miao Xie 已提交
5565
		return btrfs_ioctl_scrub(file, argp);
J
Jan Schmidt 已提交
5566
	case BTRFS_IOC_SCRUB_CANCEL:
5567
		return btrfs_ioctl_scrub_cancel(fs_info);
J
Jan Schmidt 已提交
5568
	case BTRFS_IOC_SCRUB_PROGRESS:
5569
		return btrfs_ioctl_scrub_progress(fs_info, argp);
5570
	case BTRFS_IOC_BALANCE_V2:
5571
		return btrfs_ioctl_balance(file, argp);
5572
	case BTRFS_IOC_BALANCE_CTL:
5573
		return btrfs_ioctl_balance_ctl(fs_info, arg);
5574
	case BTRFS_IOC_BALANCE_PROGRESS:
5575
		return btrfs_ioctl_balance_progress(fs_info, argp);
5576 5577
	case BTRFS_IOC_SET_RECEIVED_SUBVOL:
		return btrfs_ioctl_set_received_subvol(file, argp);
5578 5579 5580 5581
#ifdef CONFIG_64BIT
	case BTRFS_IOC_SET_RECEIVED_SUBVOL_32:
		return btrfs_ioctl_set_received_subvol_32(file, argp);
#endif
5582 5583
	case BTRFS_IOC_SEND:
		return btrfs_ioctl_send(file, argp);
5584
	case BTRFS_IOC_GET_DEV_STATS:
5585
		return btrfs_ioctl_get_dev_stats(fs_info, argp);
A
Arne Jansen 已提交
5586
	case BTRFS_IOC_QUOTA_CTL:
5587
		return btrfs_ioctl_quota_ctl(file, argp);
A
Arne Jansen 已提交
5588
	case BTRFS_IOC_QGROUP_ASSIGN:
5589
		return btrfs_ioctl_qgroup_assign(file, argp);
A
Arne Jansen 已提交
5590
	case BTRFS_IOC_QGROUP_CREATE:
5591
		return btrfs_ioctl_qgroup_create(file, argp);
A
Arne Jansen 已提交
5592
	case BTRFS_IOC_QGROUP_LIMIT:
5593
		return btrfs_ioctl_qgroup_limit(file, argp);
J
Jan Schmidt 已提交
5594 5595 5596 5597
	case BTRFS_IOC_QUOTA_RESCAN:
		return btrfs_ioctl_quota_rescan(file, argp);
	case BTRFS_IOC_QUOTA_RESCAN_STATUS:
		return btrfs_ioctl_quota_rescan_status(file, argp);
5598 5599
	case BTRFS_IOC_QUOTA_RESCAN_WAIT:
		return btrfs_ioctl_quota_rescan_wait(file, argp);
5600
	case BTRFS_IOC_DEV_REPLACE:
5601
		return btrfs_ioctl_dev_replace(fs_info, argp);
5602 5603
	case BTRFS_IOC_GET_FSLABEL:
		return btrfs_ioctl_get_fslabel(file, argp);
5604 5605
	case BTRFS_IOC_SET_FSLABEL:
		return btrfs_ioctl_set_fslabel(file, argp);
5606
	case BTRFS_IOC_GET_SUPPORTED_FEATURES:
5607
		return btrfs_ioctl_get_supported_features(argp);
5608 5609 5610 5611
	case BTRFS_IOC_GET_FEATURES:
		return btrfs_ioctl_get_features(file, argp);
	case BTRFS_IOC_SET_FEATURES:
		return btrfs_ioctl_set_features(file, argp);
C
Christoph Hellwig 已提交
5612 5613 5614 5615
	}

	return -ENOTTY;
}
5616 5617 5618 5619

#ifdef CONFIG_COMPAT
long btrfs_compat_ioctl(struct file *file, unsigned int cmd, unsigned long arg)
{
5620 5621 5622 5623
	/*
	 * These all access 32-bit values anyway so no further
	 * handling is necessary.
	 */
5624 5625 5626 5627 5628 5629 5630 5631 5632 5633 5634 5635 5636 5637 5638
	switch (cmd) {
	case FS_IOC32_GETFLAGS:
		cmd = FS_IOC_GETFLAGS;
		break;
	case FS_IOC32_SETFLAGS:
		cmd = FS_IOC_SETFLAGS;
		break;
	case FS_IOC32_GETVERSION:
		cmd = FS_IOC_GETVERSION;
		break;
	}

	return btrfs_ioctl(file, cmd, (unsigned long) compat_ptr(arg));
}
#endif